Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques


Challenge results

Task description

The goal of this task is to identify whether a machine is normal or anomalous using only normal sound data under domain shifted conditions. The main difference from DCASE 2021 Task 2 is that the domain (source domain / target domain ) of data are not available during evaluation. Therefore, the participants are expected to develop domain generalization techniques in which the output anomaly scores are not affected by the domain shifts.

More detailed task description can be found in the task description page

Teams ranking

Table including only the best performing system per submitting team.

Rank Submission Information Evaluation Dataset Development Dataset
Submission Code Technical
Report
Official
Rank
Official
Score
ToyCar
(AUC)
ToyCar
(pAUC)
ToyTrain
(AUC)
ToyTrain
(pAUC)
Fan
(AUC)
Fan
(pAUC)
Gearbox
(AUC)
Gearbox
(pAUC)
Bearing
(AUC)
Bearing
(pAUC)
Slider
(AUC)
Slider
(pAUC)
Valve
(AUC)
Valve
(pAUC)
ToyCar
(AUC)
ToyCar
(pAUC)
ToyTrain
(AUC)
ToyTrain
(pAUC)
Fan
(AUC)
Fan
(pAUC)
Gearbox
(AUC)
Gearbox
(pAUC)
Bearing
(AUC)
Bearing
(pAUC)
Slider
(AUC)
Slider
(pAUC)
Valve
(AUC)
Valve
(pAUC)
DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 68 54.01722421174122 ± 0.0002954806671840702 42.79 53.44 51.22 50.98 50.34 55.22 51.34 48.49 58.23 52.16 62.42 53.07 72.77 65.16 55.54 52.27 51.57 51.51 59.48 56.89 62.70 56.03 60.25 57.14 51.69 54.67 62.14 62.41
Bai_JLESS_task2_1 BaiJLESS2022 26 63.948583467398024 ± 0.0003628043584427405 92.17 76.25 51.78 51.22 46.52 50.73 71.80 62.91 64.02 55.80 77.31 65.19 83.15 68.73 82.50 64.68 64.52 55.60 67.28 67.25 86.77 72.67 79.82 61.04 86.73 71.41 91.63 80.01
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 5 68.22312028564296 ± 0.0003608131407524257 81.67 72.16 55.02 51.78 57.19 53.02 84.51 65.10 72.30 57.90 80.62 58.09 91.01 83.23 79.50 63.49 69.20 51.90 68.27 65.16 81.25 67.65 77.81 63.87 87.41 73.64 91.33 90.05
LEE_KNU_task2_1 LEEKNU2022 82 49.26732013587548 ± 0.00023044907066044446 53.35 51.66 45.06 49.79 49.71 50.27 49.71 49.50 47.57 49.93 47.46 49.28 49.32 51.78 89.38 75.09 79.74 63.55 90.64 78.73 91.45 80.92 95.40 75.02 88.07 75.06 87.40 75.02
Narita_AIT_task2_2 NaritaAIT2022 69 53.84966232209122 ± 0.000292419115994451 60.99 56.58 51.03 50.39 58.25 54.56 53.37 54.03 56.33 50.51 49.68 51.77 52.31 53.47 81.44 61.58 62.88 60.51 34.90 61.65 81.74 61.67 77.33 72.39 80.39 68.05 74.19 60.73
Du_NERCSLIP_task2_1 DuNERCSLIP2022 33 63.30309472782353 ± 0.0002976991866654528 63.81 59.09 58.13 51.68 53.00 51.60 82.48 60.29 67.11 54.22 67.35 56.49 87.17 75.85 80.09 63.52 69.06 58.73 90.34 80.25 86.70 66.71 86.66 67.75 92.16 83.38 90.62 83.29
Jinhyuk_SNU_task2_1 JinhyukSNU2022 78 51.02701922567823 ± 0.00029370058470223036 51.58 53.99 42.74 49.03 41.33 50.88 58.43 51.27 59.48 53.77 58.51 53.66 49.23 50.97 49.15 45.50 44.26 43.59 55.14 54.69 56.00 56.00 47.30 47.28 45.05 43.94 47.34 47.91
Hu_NJU_task2_1 HuNJU2022 65 55.07461026550378 ± 0.0003040740154131298 69.31 56.58 50.23 50.99 49.11 51.18 63.70 54.36 62.48 56.76 59.69 52.65 46.49 49.66 76.28 55.98 65.55 53.96 59.98 56.22 70.49 60.50 57.48 55.01 81.11 64.49 62.77 57.83
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 28 63.724437030531156 ± 0.000372854346756348 79.69 65.59 61.14 56.16 40.37 48.88 82.74 58.66 75.91 65.40 67.10 55.12 84.54 70.37 79.60 63.99 73.80 55.77 88.48 78.88 86.79 68.64 74.10 63.38 90.48 75.81 86.58 80.34
Wei_HEU_task2_2 WeiHEU2022 14 67.12235197698574 ± 0.0003768613481780468 97.17 88.16 57.08 53.13 49.46 51.72 85.17 60.84 70.46 58.32 74.39 60.99 84.47 67.11 82.75 60.14 69.05 59.37 69.70 59.59 81.02 65.29 69.47 54.25 89.71 76.08 93.28 72.92
Guan_HEU_task2_4 GuanHEU2022 6 68.0372804969291 ± 0.00039326172923566144 90.90 78.42 61.66 55.16 55.35 53.72 81.56 58.08 67.23 52.71 75.15 60.63 89.29 79.23 84.12 63.48 71.06 60.23 77.50 65.93 84.13 66.68 77.14 66.55 90.63 78.20 93.20 72.46
Li_CTRI_task2_1 LiCTRI2022 59 58.12785559862876 ± 0.0002695870626290191 75.84 59.32 48.76 49.93 40.15 50.15 69.07 53.18 63.30 51.01 64.95 51.74 81.71 73.11 80.35 57.21 72.38 52.78 65.49 61.14 77.96 64.19 66.88 56.04 84.48 69.36 73.16 60.35
Morita_SECOM_task2_1 MoritaSECOM2022 15 66.82507994516816 ± 0.0003343147179148136 92.52 74.39 58.30 54.78 50.27 52.52 79.90 59.71 74.06 63.18 68.54 62.22 84.96 69.42 89.55 71.91 68.58 50.36 78.26 63.83 85.07 75.64 64.44 58.21 90.82 72.99 94.05 90.64
Yamashita_GU_task2_3 YamashitaGU2022 39 62.387167480540676 ± 0.00028161700622855696 76.86 68.41 55.68 49.19 46.68 49.88 75.61 59.78 62.97 52.53 69.83 57.82 86.10 73.07 80.25 54.49 77.16 54.19 66.27 56.57 76.52 61.55 62.94 55.89 77.32 65.36 83.92 75.23
CHO_SG_task2_1 CHOSG2022 67 54.468778744613665 ± 0.0003074487298355346 72.27 56.57 54.74 51.63 49.55 50.78 63.57 55.41 52.65 51.27 47.70 50.74 53.74 54.42 46.56 49.26 56.16 52.77 47.20 53.47 59.80 55.65 56.00 52.87 74.58 63.16 61.77 60.39
Li_JAIST_task2 LiJAIST2022 83 48.93884539917759 ± 0.0002825946078999199 58.61 56.48 47.77 50.47 45.13 50.09 44.38 48.53 43.61 50.33 32.41 47.97 82.78 75.78 54.06 50.35 43.38 49.70 59.67 54.43 53.31 52.48 62.09 55.29 44.45 49.80 80.90 74.47
Gou_UESTC_task2_4 GouUESTC2022 51 59.16904508333896 ± 0.0002810411852165545 68.89 65.40 51.23 50.50 43.26 49.41 78.54 65.82 57.49 52.30 60.97 53.67 81.11 68.44 70.96 62.07 55.03 53.49 70.17 69.61 81.69 63.25 66.77 57.34 68.61 57.60 75.07 67.91
PENG_NJUPT_task2_1 PENGNJUPT2022 57 58.47341574821306 ± 0.0002990941526070191 57.29 57.37 48.52 51.03 51.04 51.16 58.64 51.13 62.47 54.62 75.49 58.51 78.03 71.20 55.55 51.02 56.84 51.95 59.26 56.68 56.38 54.52 67.42 59.08 84.36 63.45 80.88 72.00
Nejjar_ETH_task2_1 NejjarETH2022 32 63.34598205922427 ± 0.00033579696860218286 89.62 79.68 61.70 54.23 48.90 50.76 71.71 59.57 68.94 58.63 60.14 52.73 76.28 62.45 87.50 68.34 64.16 53.89 80.06 70.76 85.85 71.23 74.91 64.75 80.93 68.92 83.53 71.66
Verbitskiy_DS_task2_1 VerbitskiyDS2022 27 63.83127217235942 ± 0.0003519023225400419 71.62 65.38 52.16 51.01 44.81 50.95 86.45 74.88 65.44 57.07 72.53 57.33 92.60 80.80 75.01 59.82 61.47 52.03 64.37 66.88 69.97 59.86 80.91 69.41 81.78 63.58 89.62 79.29
Cohen_Technion_task2_2 CohenTechnion2022 63 56.763333657296656 ± 0.0002975695814466658 75.61 65.47 50.50 50.02 43.47 50.14 67.34 51.86 64.40 58.88 63.78 53.55 55.09 51.75 77.70 55.22 55.96 49.89 62.80 55.19 69.72 56.91 59.51 50.99 80.22 59.78 54.43 50.34
Deng_THU_task2_4 DengTHU2022 7 67.62358685887781 ± 0.00038777202086470904 88.48 70.07 60.00 53.70 44.48 50.01 85.33 66.41 71.30 60.33 81.55 67.04 89.68 84.10 87.09 66.28 81.47 69.14 88.07 77.00 89.11 75.58 87.21 75.10 91.94 84.87 98.17 94.83
Liu_BUPT_task2_2 LiuBUPT2022 49 59.767720255838206 ± 0.00027644210188000623 53.83 57.79 55.34 51.27 43.14 50.25 69.68 51.03 69.97 57.28 71.37 59.82 85.14 80.49 78.48 56.66 62.40 54.65 71.34 58.11 76.72 59.61 68.31 56.65 88.28 73.78 83.97 70.11
Kazakova_ITMO_task2_1 KazakovaITMO2022 80 49.89612147323122 ± 0.0002773039817690667 58.35 54.92 51.25 50.68 48.80 50.47 52.70 49.56 39.56 49.89 39.76 49.96 62.81 55.73 51.56 53.23 52.11 52.84 57.85 54.60 50.13 53.15 47.47 53.81 50.10 56.40 17.68 49.27
Kodua_ITMO_task2_1 KoduaITMO2022 74 52.96330674177412 ± 0.00029557613542138784 53.66 51.39 48.84 49.30 41.08 48.93 59.15 51.66 56.48 54.50 63.19 52.88 61.24 51.89 56.25 51.75 55.47 52.13 59.31 52.22 61.08 53.33 55.48 51.47 68.85 56.83 59.08 50.82
Liu_CQUPT_task2_4 LiuCQUPT2022 1 70.97462888079512 ± 0.0003427217026837889 88.45 81.83 70.46 61.14 57.34 57.33 86.04 64.22 68.85 54.45 78.26 66.39 83.87 75.22 81.50 67.12 76.22 64.27 70.11 65.86 89.28 76.04 81.96 70.89 90.08 79.80 93.84 86.80
Siang_NTHU_task2_1 SiangNTHU2022 79 50.59102827288451 ± 0.00027303186490778593 57.84 52.44 44.31 49.56 46.32 50.42 53.15 50.25 38.27 50.91 58.50 51.98 60.33 58.42 78.39 61.24 64.60 55.08 70.73 61.34 82.91 65.22 75.63 65.16 89.87 76.54 88.56 79.06
Tozicka_NSW_task2_1 TozickaNSW2022 38 62.50004793576084 ± 0.00034772455431082163 88.28 74.95 58.44 51.01 42.75 52.14 87.23 72.17 60.35 55.15 62.65 52.16 74.38 70.55 88.24 67.80 72.19 63.80 63.51 60.40 80.22 64.80 70.66 61.00 78.06 62.60 88.73 76.20
Almudevar_UZ_task2_2 AlmudevarUZ2022 45 60.60499062666268 ± 0.00031377932481308624 89.58 77.07 47.34 49.83 47.90 52.47 75.43 56.23 64.84 54.36 68.26 55.78 67.92 55.97 82.62 56.20 60.16 51.18 70.45 60.20 73.88 55.34 63.86 59.15 89.83 77.74 61.77 53.07
Jalalia_AIT_task2_1 JalaliaAIT2022 77 51.338525836051105 ± 0.00029471041821909645 51.35 55.59 44.78 49.55 41.06 50.21 58.20 51.69 60.13 55.52 58.51 53.13 48.68 51.11 63.47 52.75 52.36 50.68 59.38 57.85 63.91 57.90 54.11 51.72 58.28 55.23 48.69 50.48
Zorin_AIRI_task2_1 ZorinAIRI2022 84 43.16915219364365 ± 0.0002620788289881202 37.17 49.65 45.02 49.74 43.23 49.83 40.98 49.49 38.40 50.13 40.83 48.68 39.49 48.92 54.64 52.77 60.43 51.17 54.28 50.10 56.27 50.32 46.42 48.66 63.60 53.43 47.75 51.18
Venkatesh_MERL_task2_3 VenkateshMERL2022 8 67.56529995172603 ± 0.000353158822539437 93.88 78.67 58.23 54.73 48.17 50.34 86.76 79.43 72.54 61.86 73.64 60.70 83.72 62.93 83.66 65.19 66.38 57.27 64.72 61.61 87.47 70.70 73.16 57.59 90.16 77.60 97.35 90.76


Supplementary metrics (recall, precision, and F1 score)

Rank Submission Information Evaluation Dataset
Submission Code Technical
Report
Official
Rank
ToyCar
(F1 score)
ToyCar
(Recall)
ToyCar
(Precision)
ToyTrain
(F1 score)
ToyTrain
(Recall)
ToyTrain
(Precision)
Fan
(F1 score)
Fan
(Recall)
Fan
(Precision)
Gearbox
(F1 score)
Gearbox
(Recall)
Gearbox
(Precision)
Bearing
(F1 score)
Bearing
(Recall)
Bearing
(Precision)
Slider
(F1 score)
Slider
(Recall)
Slider
(Precision)
Valve
(F1 score)
Valve
(Recall)
Valve
(Precision)
DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 68 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 55.14 52.94 57.52 43.31 34.77 57.41 42.52 31.43 65.69 0.00 0.00 0.00
Bai_JLESS_task2_1 BaiJLESS2022 26 85.28 94.80 77.50 0.00 0.00 0.00 25.73 16.50 58.35 66.70 62.01 72.15 64.14 71.76 57.98 71.97 80.66 64.97 61.21 49.20 80.99
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 5 75.29 74.95 75.63 52.38 49.58 55.52 54.05 49.01 60.26 80.57 80.34 80.81 68.50 68.20 68.80 76.59 76.34 76.84 82.14 80.83 83.49
LEE_KNU_task2_1 LEEKNU2022 82 57.67 65.42 51.56 32.71 26.10 43.79 60.10 72.26 51.44 50.39 50.78 50.00 60.40 79.18 48.82 47.79 46.75 48.88 50.26 49.83 50.70
Narita_AIT_task2_2 NaritaAIT2022 69 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00
Du_NERCSLIP_task2_1 DuNERCSLIP2022 33 64.19 86.32 51.10 66.74 99.32 50.25 63.40 83.11 51.25 77.05 87.94 68.56 69.12 97.30 53.60 50.17 39.31 69.29 0.00 0.00 0.00
Jinhyuk_SNU_task2_1 JinhyukSNU2022 78 41.62 32.89 56.66 66.67 100.00 50.00 0.00 0.00 0.00 69.04 91.91 55.29 62.33 75.94 52.85 47.11 39.15 59.14 21.80 13.76 52.41
Hu_NJU_task2_1 HuNJU2022 65 0.00 0.00 0.00 0.00 0.00 0.00 13.47 7.66 55.52 0.00 0.00 0.00 0.00 0.00 0.00 51.63 45.59 59.51 0.00 0.00 0.00
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 28 67.19 100.00 50.59 67.57 98.18 51.51 56.01 63.98 49.81 66.82 100.00 50.17 67.11 100.00 50.51 68.61 99.31 52.41 74.10 98.95 59.22
Wei_HEU_task2_2 WeiHEU2022 14 85.41 78.80 93.24 50.75 46.23 56.26 36.26 26.45 57.65 78.80 76.60 81.13 61.32 53.04 72.66 67.36 63.28 71.99 70.96 67.71 74.54
Guan_HEU_task2_4 GuanHEU2022 6 78.27 93.46 67.33 55.12 47.75 65.18 54.13 56.84 51.67 72.66 95.58 58.60 16.21 9.34 61.31 64.39 65.53 63.29 78.75 92.22 68.71
Li_CTRI_task2_1 LiCTRI2022 59 69.87 74.12 66.09 39.57 34.10 47.12 0.00 0.00 0.00 54.27 43.07 73.34 67.26 80.16 57.94 29.58 19.25 63.87 0.00 0.00 0.00
Morita_SECOM_task2_1 MoritaSECOM2022 15 69.69 56.47 90.99 0.00 0.00 0.00 15.71 9.09 57.80 42.25 29.92 71.86 0.00 0.00 0.00 67.97 64.01 72.44 78.36 82.04 75.00
Yamashita_GU_task2_3 YamashitaGU2022 39 61.39 49.75 80.16 44.19 34.76 60.63 35.00 26.11 53.10 63.93 56.20 74.13 57.97 50.23 68.55 60.61 53.71 69.55 67.46 55.67 85.60
CHO_SG_task2_1 CHOSG2022 67 69.65 84.99 59.01 66.33 90.90 52.21 63.47 87.45 49.81 66.56 96.20 50.89 0.00 0.00 0.00 62.14 85.54 48.79 65.45 93.44 50.37
Li_JAIST_task2 LiJAIST2022 83 66.58 96.96 50.69 0.00 0.00 0.00 0.00 0.00 0.00 14.90 9.71 32.03 64.53 92.72 49.48 0.00 0.00 0.00 30.99 19.20 80.23
Gou_UESTC_task2_4 GouUESTC2022 51 63.10 62.91 63.29 46.35 42.30 51.26 45.61 43.84 47.53 72.48 72.02 72.96 54.16 52.31 56.14 65.07 72.26 59.18 71.51 70.75 72.29
PENG_NJUPT_task2_1 PENGNJUPT2022 57 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 54.22 51.13 57.70 67.33 78.30 59.06 76.81 90.14 66.92 65.46 57.15 76.59
Nejjar_ETH_task2_1 NejjarETH2022 32 72.84 98.95 57.63 40.59 32.74 53.40 0.00 0.00 0.00 64.47 81.18 53.47 66.74 100.00 50.08 66.67 100.00 50.00 37.11 26.87 60.00
Verbitskiy_DS_task2_1 VerbitskiyDS2022 27 66.08 72.32 60.83 56.54 64.15 50.55 21.91 13.70 54.72 78.58 79.34 77.84 60.34 57.99 62.89 65.83 59.64 73.45 83.13 82.92 83.34
Cohen_Technion_task2_2 CohenTechnion2022 63 62.28 58.49 66.58 49.46 46.88 52.33 45.49 41.23 50.74 64.84 64.57 65.11 59.83 58.20 61.55 60.24 58.87 61.68 54.21 54.14 54.27
Deng_THU_task2_4 DengTHU2022 7 67.64 100.00 51.11 37.00 28.94 51.28 43.49 38.91 49.30 70.70 95.25 56.21 66.55 98.57 50.24 68.75 98.24 52.88 77.22 94.72 65.17
Liu_BUPT_task2_2 LiuBUPT2022 49 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 14.99 8.57 59.60 8.52 4.55 66.09 0.00 0.00 0.00 0.00 0.00 0.00
Kazakova_ITMO_task2_1 KazakovaITMO2022 80 64.79 92.43 49.87 63.01 79.09 52.37 55.94 63.85 49.77 66.70 84.90 54.93 53.50 68.94 43.71 49.85 60.32 42.48 63.49 75.63 54.71
Kodua_ITMO_task2_1 KoduaITMO2022 74 7.02 3.75 55.05 0.00 0.00 0.00 0.00 0.00 0.00 20.07 11.93 63.05 16.27 9.12 75.43 0.00 0.00 0.00 0.00 0.00 0.00
Liu_CQUPT_task2_4 LiuCQUPT2022 1 77.69 75.86 79.62 64.69 71.15 59.30 45.11 37.68 56.19 79.35 78.82 79.89 65.25 63.81 66.75 70.52 68.76 72.38 71.03 68.47 73.80
Siang_NTHU_task2_1 SiangNTHU2022 79 66.74 100.00 50.08 65.15 95.93 49.33 56.77 67.61 48.93 57.32 59.96 54.90 60.10 77.17 49.22 58.21 65.32 52.50 66.03 93.61 51.00
Tozicka_NSW_task2_1 TozickaNSW2022 38 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Almudevar_UZ_task2_2 AlmudevarUZ2022 45 79.18 76.79 81.72 48.67 45.94 51.74 46.85 43.62 50.60 71.63 71.16 72.09 61.79 58.21 65.85 64.93 64.03 65.86 62.56 61.63 63.53
Jalalia_AIT_task2_1 JalaliaAIT2022 77 36.45 27.61 53.63 0.00 0.00 0.00 14.09 8.37 44.56 37.97 30.24 51.03 59.71 67.08 53.79 27.09 16.71 71.46 10.14 5.60 53.76
Zorin_AIRI_task2_1 ZorinAIRI2022 84 7.46 3.89 88.89 0.00 0.00 0.00 12.32 7.09 46.88 4.05 2.18 27.91 12.78 7.21 56.46 0.00 0.00 0.00 6.48 3.48 47.06
Venkatesh_MERL_task2_3 VenkateshMERL2022 8 81.19 100.00 68.34 42.50 31.38 65.84 58.09 69.75 49.78 0.00 0.00 0.00 67.10 96.92 51.31 67.88 77.42 60.43 75.78 82.49 70.08



Systems ranking

Rank Submission Information Evaluation Dataset Development Dataset
Submission Code Technical
Report
Official
Rank
Official
Score
ToyCar
(AUC)
ToyCar
(pAUC)
ToyTrain
(AUC)
ToyTrain
(pAUC)
Fan
(AUC)
Fan
(pAUC)
Gearbox
(AUC)
Gearbox
(pAUC)
Bearing
(AUC)
Bearing
(pAUC)
Slider
(AUC)
Slider
(pAUC)
Valve
(AUC)
Valve
(pAUC)
ToyCar
(AUC)
ToyCar
(pAUC)
ToyTrain
(AUC)
ToyTrain
(pAUC)
Fan
(AUC)
Fan
(pAUC)
Gearbox
(AUC)
Gearbox
(pAUC)
Bearing
(AUC)
Bearing
(pAUC)
Slider
(AUC)
Slider
(pAUC)
Valve
(AUC)
Valve
(pAUC)
DCASE2022_baseline_task2_AE DCASE2022baseline2022 75 52.941609218413156 ± 0.0002952567128951358 61.18 60.21 43.14 49.36 41.16 50.12 61.92 51.95 59.93 53.95 58.95 54.16 54.26 51.30 62.61 52.74 49.83 50.48 62.89 57.52 65.78 58.49 56.40 51.98 62.81 55.78 50.73 50.36
DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 68 54.01722421174122 ± 0.0002954806671840702 42.79 53.44 51.22 50.98 50.34 55.22 51.34 48.49 58.23 52.16 62.42 53.07 72.77 65.16 55.54 52.27 51.57 51.51 59.48 56.89 62.70 56.03 60.25 57.14 51.69 54.67 62.14 62.41
Bai_JLESS_task2_1 BaiJLESS2022 26 63.948583467398024 ± 0.0003628043584427405 92.17 76.25 51.78 51.22 46.52 50.73 71.80 62.91 64.02 55.80 77.31 65.19 83.15 68.73 82.50 64.68 64.52 55.60 67.28 67.25 86.77 72.67 79.82 61.04 86.73 71.41 91.63 80.01
Bai_JLESS_task2_2 BaiJLESS2022 44 61.101918636073826 ± 0.0003308162422994587 53.39 53.29 50.72 52.00 52.82 55.74 72.85 66.55 64.93 55.63 77.31 65.19 78.15 66.94 73.40 62.27 62.12 51.43 60.61 65.86 85.53 71.33 72.70 56.38 86.73 71.41 87.33 79.32
Bai_JLESS_task2_3 BaiJLESS2022 42 62.207778577741045 ± 0.00032278441727734837 92.17 76.25 51.78 51.22 46.52 50.73 71.14 50.61 68.22 55.34 66.64 55.43 83.15 68.73 82.50 64.68 64.52 55.60 67.28 67.25 67.00 60.46 73.91 69.92 82.80 68.20 91.63 80.01
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2022 20 66.5026110256853 ± 0.0003508467181733295 65.39 71.51 56.62 51.92 48.94 52.66 82.20 64.80 78.56 61.55 80.79 58.82 92.48 83.89 74.62 61.57 62.71 50.69 66.86 64.67 80.98 66.27 82.00 65.76 82.24 70.71 86.50 89.94
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 5 68.22312028564296 ± 0.0003608131407524257 81.67 72.16 55.02 51.78 57.19 53.02 84.51 65.10 72.30 57.90 80.62 58.09 91.01 83.23 79.50 63.49 69.20 51.90 68.27 65.16 81.25 67.65 77.81 63.87 87.41 73.64 91.33 90.05
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2022 52 59.034000365177555 ± 0.0003364721172887524 76.33 71.84 55.39 52.77 58.59 54.80 87.93 68.57 73.72 56.55 26.82 56.92 90.70 84.45 87.99 71.32 74.90 58.75 80.02 72.53 91.62 76.43 91.66 82.19 93.41 87.71 95.39 93.31
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2022 12 67.14222981034025 ± 0.00036474217704024595 85.61 73.71 45.81 51.42 53.79 55.96 83.67 63.00 76.85 56.89 81.34 59.82 92.46 87.91 91.46 71.20 79.84 57.05 83.97 74.89 92.16 78.24 93.35 86.65 95.25 86.04 97.63 94.83
LEE_KNU_task2_1 LEEKNU2022 82 49.26732013587548 ± 0.00023044907066044446 53.35 51.66 45.06 49.79 49.71 50.27 49.71 49.50 47.57 49.93 47.46 49.28 49.32 51.78 89.38 75.09 79.74 63.55 90.64 78.73 91.45 80.92 95.40 75.02 88.07 75.06 87.40 75.02
Narita_AIT_task2_1 NaritaAIT2022 70 53.51232629097276 ± 0.0002912597694015261 58.99 53.44 47.73 51.05 59.56 54.92 56.28 55.62 55.42 49.76 48.62 51.04 51.32 58.32 48.59 53.21 55.64 50.69 35.43 61.81 86.03 63.17 74.29 70.92 81.00 69.67 85.19 75.78
Narita_AIT_task2_2 NaritaAIT2022 69 53.84966232209122 ± 0.000292419115994451 60.99 56.58 51.03 50.39 58.25 54.56 53.37 54.03 56.33 50.51 49.68 51.77 52.31 53.47 81.44 61.58 62.88 60.51 34.90 61.65 81.74 61.67 77.33 72.39 80.39 68.05 74.19 60.73
Narita_AIT_task2_3 NaritaAIT2022 76 52.358564643955404 ± 0.00028732610320381395 57.63 52.21 49.37 50.53 54.55 52.98 50.82 52.00 55.78 51.06 50.64 50.94 50.04 54.77 69.32 62.21 63.99 56.50 75.35 72.01 86.58 74.05 53.95 63.92 88.66 77.59 68.04 58.23
Narita_AIT_task2_4 NaritaAIT2022 72 53.27439248823605 ± 0.0002960338887133308 60.99 56.58 51.03 50.39 54.55 52.98 50.82 52.00 56.33 50.51 50.64 50.94 51.32 58.32 81.44 61.58 62.88 60.51 75.35 72.01 86.58 74.05 77.33 72.39 88.66 77.59 85.19 75.78
Du_NERCSLIP_task2_1 DuNERCSLIP2022 33 63.30309472782353 ± 0.0002976991866654528 63.81 59.09 58.13 51.68 53.00 51.60 82.48 60.29 67.11 54.22 67.35 56.49 87.17 75.85 80.09 63.52 69.06 58.73 90.34 80.25 86.70 66.71 86.66 67.75 92.16 83.38 90.62 83.29
Du_NERCSLIP_task2_2 DuNERCSLIP2022 37 62.579607866831424 ± 0.0003032188907543935 56.08 57.79 59.10 52.34 53.00 51.60 81.70 60.78 67.11 54.22 67.61 54.95 87.78 78.13 78.19 62.00 69.30 59.41 90.34 80.25 85.25 67.03 86.66 67.75 92.18 82.76 90.48 82.95
Du_NERCSLIP_task2_3 DuNERCSLIP2022 35 62.78654798621621 ± 0.00029667454336750667 56.08 57.79 59.36 52.93 53.73 51.49 82.02 60.27 68.15 53.81 67.61 54.95 87.78 78.13 78.19 62.00 68.70 58.52 90.25 79.99 85.25 67.03 86.66 67.75 92.18 82.76 90.48 82.95
Du_NERCSLIP_task2_4 DuNERCSLIP2022 36 62.74927690762567 ± 0.0003068930687963723 51.45 57.36 58.52 53.67 56.65 52.70 80.77 58.61 68.27 51.92 71.64 55.84 88.56 80.11 76.31 61.25 68.59 58.94 90.44 80.15 84.09 66.65 86.01 67.88 91.97 80.86 90.06 82.94
Jinhyuk_SNU_task2_1 JinhyukSNU2022 78 51.02701922567823 ± 0.00029370058470223036 51.58 53.99 42.74 49.03 41.33 50.88 58.43 51.27 59.48 53.77 58.51 53.66 49.23 50.97 49.15 45.50 44.26 43.59 55.14 54.69 56.00 56.00 47.30 47.28 45.05 43.94 47.34 47.91
Hu_NJU_task2_1 HuNJU2022 65 55.07461026550378 ± 0.0003040740154131298 69.31 56.58 50.23 50.99 49.11 51.18 63.70 54.36 62.48 56.76 59.69 52.65 46.49 49.66 76.28 55.98 65.55 53.96 59.98 56.22 70.49 60.50 57.48 55.01 81.11 64.49 62.77 57.83
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 28 63.724437030531156 ± 0.000372854346756348 79.69 65.59 61.14 56.16 40.37 48.88 82.74 58.66 75.91 65.40 67.10 55.12 84.54 70.37 79.60 63.99 73.80 55.77 88.48 78.88 86.79 68.64 74.10 63.38 90.48 75.81 86.58 80.34
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2022 30 63.55709677474374 ± 0.000361118136614277 78.05 65.44 60.99 56.58 41.85 48.82 80.53 58.69 74.77 64.43 67.68 54.78 82.31 68.87 79.14 61.97 74.46 54.54 88.22 76.90 87.21 70.09 73.00 62.40 89.94 74.48 85.38 78.03
Wei_HEU_task2_1 WeiHEU2022 29 63.603340574118775 ± 0.0003498881455140153 72.75 61.29 62.95 53.68 44.98 51.64 79.07 57.22 67.62 54.69 72.24 55.56 85.74 80.03 48.62 49.91 52.09 49.62 64.42 60.95 75.97 64.21 78.58 64.28 79.45 63.22 70.80 63.32
Wei_HEU_task2_2 WeiHEU2022 14 67.12235197698574 ± 0.0003768613481780468 97.17 88.16 57.08 53.13 49.46 51.72 85.17 60.84 70.46 58.32 74.39 60.99 84.47 67.11 82.75 60.14 69.05 59.37 69.70 59.59 81.02 65.29 69.47 54.25 89.71 76.08 93.28 72.92
Wei_HEU_task2_3 WeiHEU2022 17 66.65219954899092 ± 0.0003672678022080277 95.88 87.26 57.08 53.13 49.05 50.06 85.26 62.20 67.62 54.69 74.59 60.55 86.27 70.31 83.79 61.99 69.05 59.37 74.16 63.39 82.34 66.99 78.58 64.28 90.05 77.09 93.56 74.19
Guan_HEU_task2_1 GuanHEU2022 22 66.3946192973576 ± 0.0003745835334668857 89.11 78.69 58.18 53.09 49.20 51.73 83.53 63.01 71.70 58.95 74.30 60.35 82.27 64.62 80.65 58.12 67.81 59.07 71.44 62.13 80.25 64.69 69.41 54.74 87.96 73.33 92.61 70.23
Guan_HEU_task2_2 GuanHEU2022 19 66.53202926501359 ± 0.00038128464064027007 94.13 85.69 58.18 53.11 49.20 51.73 83.89 60.90 71.51 57.88 73.59 60.27 82.13 62.86 84.69 60.55 68.02 59.69 71.44 62.13 82.22 65.85 70.80 54.22 89.91 76.74 91.68 66.02
Guan_HEU_task2_3 GuanHEU2022 11 67.41857863156892 ± 0.00036908458449298254 85.05 70.92 61.82 55.11 55.35 53.72 81.40 58.39 67.20 52.99 75.19 61.29 88.22 78.67 82.12 63.07 71.03 59.99 77.50 65.93 83.43 64.82 77.08 67.03 89.50 74.94 93.43 73.48
Guan_HEU_task2_4 GuanHEU2022 6 68.0372804969291 ± 0.00039326172923566144 90.90 78.42 61.66 55.16 55.35 53.72 81.56 58.08 67.23 52.71 75.15 60.63 89.29 79.23 84.12 63.48 71.06 60.23 77.50 65.93 84.13 66.68 77.14 66.55 90.63 78.20 93.20 72.46
Li_CTRI_task2_1 LiCTRI2022 59 58.12785559862876 ± 0.0002695870626290191 75.84 59.32 48.76 49.93 40.15 50.15 69.07 53.18 63.30 51.01 64.95 51.74 81.71 73.11 80.35 57.21 72.38 52.78 65.49 61.14 77.96 64.19 66.88 56.04 84.48 69.36 73.16 60.35
Morita_SECOM_task2_1 MoritaSECOM2022 15 66.82507994516816 ± 0.0003343147179148136 92.52 74.39 58.30 54.78 50.27 52.52 79.90 59.71 74.06 63.18 68.54 62.22 84.96 69.42 89.55 71.91 68.58 50.36 78.26 63.83 85.07 75.64 64.44 58.21 90.82 72.99 94.05 90.64
Morita_SECOM_task2_2 MoritaSECOM2022 25 65.0882650554134 ± 0.00035810965057196366 93.71 77.55 61.29 53.34 45.76 50.34 88.80 72.63 67.22 53.43 66.19 62.21 73.01 66.63 88.80 71.00 77.29 57.24 77.94 65.34 88.26 72.98 68.45 55.99 91.73 74.07 93.48 85.61
Morita_SECOM_task2_3 MoritaSECOM2022 21 66.39941023196549 ± 0.00034462400453683404 92.52 74.39 61.29 53.34 45.76 50.34 88.80 72.63 72.14 60.08 66.19 62.21 80.49 69.16 89.55 71.91 77.29 57.24 77.94 65.34 88.26 72.98 69.40 59.98 91.73 74.07 94.06 91.62
Morita_SECOM_task2_4 MoritaSECOM2022 16 66.72449290579897 ± 0.00030788314474219496 92.52 74.39 61.29 53.34 46.72 51.65 88.80 72.63 72.14 60.08 66.19 62.21 82.14 67.44 89.55 71.91 77.29 57.24 80.13 64.80 88.26 72.98 69.40 59.98 91.73 74.07 97.61 88.66
Yamashita_GU_task2_1 YamashitaGU2022 53 58.94125019412099 ± 0.0003148043944963588 63.39 59.68 55.68 49.19 42.27 49.61 67.60 55.88 56.07 50.26 72.37 60.04 86.10 73.07 75.81 54.58 77.16 54.19 62.56 56.32 71.25 56.75 59.74 53.68 76.90 63.60 83.92 75.23
Yamashita_GU_task2_2 YamashitaGU2022 61 57.96512591514972 ± 0.00031675252816225114 76.86 68.41 52.99 49.70 43.57 50.34 75.61 59.78 57.07 51.10 69.83 57.82 57.13 51.88 80.25 54.49 75.04 52.53 54.49 55.26 76.52 61.55 48.48 54.42 77.32 65.36 60.12 50.54
Yamashita_GU_task2_3 YamashitaGU2022 39 62.387167480540676 ± 0.00028161700622855696 76.86 68.41 55.68 49.19 46.68 49.88 75.61 59.78 62.97 52.53 69.83 57.82 86.10 73.07 80.25 54.49 77.16 54.19 66.27 56.57 76.52 61.55 62.94 55.89 77.32 65.36 83.92 75.23
Yamashita_GU_task2_4 YamashitaGU2022 48 60.209216780036876 ± 0.000283227668146682 68.38 62.67 55.46 49.28 38.98 49.71 75.61 59.78 62.19 52.26 71.92 60.20 86.10 73.07 78.98 56.69 77.16 54.26 66.62 56.09 76.52 61.55 63.01 55.85 77.78 64.36 83.92 75.23
CHO_SG_task2_1 CHOSG2022 67 54.468778744613665 ± 0.0003074487298355346 72.27 56.57 54.74 51.63 49.55 50.78 63.57 55.41 52.65 51.27 47.70 50.74 53.74 54.42 46.56 49.26 56.16 52.77 47.20 53.47 59.80 55.65 56.00 52.87 74.58 63.16 61.77 60.39
CHO_SG_task2_2 CHOSG2022 81 49.28781187514513 ± 0.00027811585615677086 49.35 51.09 49.41 50.52 46.33 50.95 49.88 51.96 50.24 50.29 45.40 50.29 50.21 49.76 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Li_JAIST_task2 LiJAIST2022 83 48.93884539917759 ± 0.0002825946078999199 58.61 56.48 47.77 50.47 45.13 50.09 44.38 48.53 43.61 50.33 32.41 47.97 82.78 75.78 54.06 50.35 43.38 49.70 59.67 54.43 53.31 52.48 62.09 55.29 44.45 49.80 80.90 74.47
Gou_UESTC_task2_1 GouUESTC2022 56 58.722651329838406 ± 0.00030480919406067754 68.89 65.40 51.23 50.50 51.62 51.44 78.54 65.82 55.59 50.83 64.37 57.99 61.39 52.48 70.96 62.07 55.03 53.49 62.90 64.45 81.69 63.25 61.21 53.49 67.00 62.02 61.60 51.96
Gou_UESTC_task2_2 GouUESTC2022 58 58.434451862538516 ± 0.0003140236598222822 61.39 58.43 48.16 49.85 49.16 52.70 71.40 62.27 57.16 51.89 60.78 55.76 81.11 68.44 53.20 52.75 53.44 52.23 61.50 64.40 62.86 54.72 57.56 51.80 68.35 58.09 75.07 67.91
Gou_UESTC_task2_3 GouUESTC2022 60 58.0929101644006 ± 0.00030998986104600966 70.88 62.96 50.24 51.22 43.26 49.41 74.65 66.89 65.83 55.69 66.71 58.58 55.65 54.02 64.85 57.15 48.53 50.94 70.17 69.61 72.80 59.44 55.12 48.91 63.30 58.38 58.17 55.50
Gou_UESTC_task2_4 GouUESTC2022 51 59.16904508333896 ± 0.0002810411852165545 68.89 65.40 51.23 50.50 43.26 49.41 78.54 65.82 57.49 52.30 60.97 53.67 81.11 68.44 70.96 62.07 55.03 53.49 70.17 69.61 81.69 63.25 66.77 57.34 68.61 57.60 75.07 67.91
PENG_NJUPT_task2_1 PENGNJUPT2022 57 58.47341574821306 ± 0.0002990941526070191 57.29 57.37 48.52 51.03 51.04 51.16 58.64 51.13 62.47 54.62 75.49 58.51 78.03 71.20 55.55 51.02 56.84 51.95 59.26 56.68 56.38 54.52 67.42 59.08 84.36 63.45 80.88 72.00
PENG_NJUPT_task2_2 PENGNJUPT2022 62 57.37883429943842 ± 0.0002972201845669457 53.23 56.22 51.88 51.59 50.03 51.08 56.69 51.14 65.22 52.09 76.71 60.96 63.61 68.86 59.28 55.92 59.53 50.84 51.48 61.75 64.38 58.02 70.55 58.48 87.44 70.20 80.25 67.41
Nejjar_ETH_task2_1 NejjarETH2022 32 63.34598205922427 ± 0.00033579696860218286 89.62 79.68 61.70 54.23 48.90 50.76 71.71 59.57 68.94 58.63 60.14 52.73 76.28 62.45 87.50 68.34 64.16 53.89 80.06 70.76 85.85 71.23 74.91 64.75 80.93 68.92 83.53 71.66
Nejjar_ETH_task2_2 NejjarETH2022 54 58.84971221021993 ± 0.00033803435798248585 89.62 79.68 61.70 54.23 30.59 48.65 71.71 59.57 68.94 58.63 60.14 52.73 76.28 62.45 87.50 68.34 64.16 53.89 80.06 70.76 85.85 71.23 74.91 64.75 80.93 68.92 83.53 71.66
Verbitskiy_DS_task2_1 VerbitskiyDS2022 27 63.83127217235942 ± 0.0003519023225400419 71.62 65.38 52.16 51.01 44.81 50.95 86.45 74.88 65.44 57.07 72.53 57.33 92.60 80.80 75.01 59.82 61.47 52.03 64.37 66.88 69.97 59.86 80.91 69.41 81.78 63.58 89.62 79.29
Verbitskiy_DS_task2_2 VerbitskiyDS2022 40 62.38199546681628 ± 0.00033876186498202777 71.62 65.38 51.94 50.99 44.81 50.95 78.84 63.40 65.44 57.07 67.00 55.58 92.60 80.80 75.01 59.82 64.25 53.23 64.37 66.88 84.85 72.82 80.91 69.41 83.10 66.38 89.62 79.29
Verbitskiy_DS_task2_3 VerbitskiyDS2022 34 62.90263916891926 ± 0.00034513471368625894 76.93 64.53 53.45 52.63 43.61 50.53 78.84 63.40 65.44 57.07 67.00 55.58 93.88 83.97 77.64 62.00 62.03 53.34 70.81 63.73 84.85 72.82 81.02 69.31 83.17 66.33 90.91 80.59
Verbitskiy_DS_task2_4 VerbitskiyDS2022 31 63.392855035610566 ± 0.0003502721194045189 78.13 64.99 54.50 52.24 43.57 50.62 78.84 63.27 67.91 57.72 67.00 55.58 93.81 85.70 77.91 62.06 64.22 54.00 70.81 63.73 84.88 72.82 82.03 68.84 83.17 66.33 91.31 83.05
Cohen_Technion_task2_1 CohenTechnion2022 64 55.64021966150642 ± 0.0002945840301639103 71.76 58.89 52.66 51.12 44.49 49.54 62.52 50.25 58.24 55.89 54.40 49.33 66.19 54.01 64.31 54.80 55.67 49.64 55.64 49.98 66.61 55.99 57.96 53.69 81.36 63.37 54.34 50.48
Cohen_Technion_task2_2 CohenTechnion2022 63 56.763333657296656 ± 0.0002975695814466658 75.61 65.47 50.50 50.02 43.47 50.14 67.34 51.86 64.40 58.88 63.78 53.55 55.09 51.75 77.70 55.22 55.96 49.89 62.80 55.19 69.72 56.91 59.51 50.99 80.22 59.78 54.43 50.34
Deng_THU_task2_1 DengTHU2022 9 67.52738146990447 ± 0.00032810483165450423 79.52 67.45 62.91 55.52 43.93 53.34 87.78 69.90 66.54 58.18 81.75 66.45 93.08 86.30 90.13 72.91 83.44 71.47 88.69 78.80 91.58 80.48 90.62 82.57 92.02 85.03 98.81 96.74
Deng_THU_task2_2 DengTHU2022 18 66.53259340769795 ± 0.0003299058303031156 80.16 68.33 61.93 55.16 40.78 52.18 87.78 69.90 66.33 58.83 80.88 64.86 93.27 86.33 90.11 72.88 83.16 70.73 88.30 77.94 91.58 80.48 90.66 82.43 91.75 83.98 98.80 96.62
Deng_THU_task2_3 DengTHU2022 13 67.13366725064618 ± 0.0003089590683108787 78.34 67.08 59.49 54.77 43.89 52.95 88.99 71.15 67.01 58.24 81.99 66.54 93.08 86.30 88.67 72.02 82.63 70.52 88.61 78.26 91.66 80.31 90.54 82.38 91.90 85.18 98.81 96.74
Deng_THU_task2_4 DengTHU2022 7 67.62358685887781 ± 0.00038777202086470904 88.48 70.07 60.00 53.70 44.48 50.01 85.33 66.41 71.30 60.33 81.55 67.04 89.68 84.10 87.09 66.28 81.47 69.14 88.07 77.00 89.11 75.58 87.21 75.10 91.94 84.87 98.17 94.83
Liu_BUPT_task2_1 LiuBUPT2022 55 58.80715749913074 ± 0.0003208251356634862 56.43 52.25 45.74 49.79 45.69 50.15 69.68 51.03 69.97 57.28 71.37 59.82 85.14 80.49 61.34 52.58 51.82 51.37 61.33 56.72 76.72 59.61 68.31 56.65 88.28 73.78 83.97 70.11
Liu_BUPT_task2_2 LiuBUPT2022 49 59.767720255838206 ± 0.00027644210188000623 53.83 57.79 55.34 51.27 43.14 50.25 69.68 51.03 69.97 57.28 71.37 59.82 85.14 80.49 78.48 56.66 62.40 54.65 71.34 58.11 76.72 59.61 68.31 56.65 88.28 73.78 83.97 70.11
Kazakova_ITMO_task2_1 KazakovaITMO2022 80 49.89612147323122 ± 0.0002773039817690667 58.35 54.92 51.25 50.68 48.80 50.47 52.70 49.56 39.56 49.89 39.76 49.96 62.81 55.73 51.56 53.23 52.11 52.84 57.85 54.60 50.13 53.15 47.47 53.81 50.10 56.40 17.68 49.27
Kazakova_ITMO_task2_2 KazakovaITMO2022 86 37.56266598852081 ± 0.0002693154734263868 31.34 47.66 49.82 50.69 43.36 50.15 52.95 51.45 48.47 52.31 45.70 52.35 13.61 47.90 52.20 50.60 51.70 51.30 55.20 50.60 64.80 51.60 50.65 50.60 57.00 53.20 54.30 52.90
Kazakova_ITMO_task2_3 KazakovaITMO2022 85 39.578820200597065 ± 0.00024604960892591796 58.35 54.92 51.25 50.68 48.80 50.47 52.95 51.45 39.56 49.89 45.70 52.35 13.61 47.90 51.56 53.23 52.11 52.84 57.85 54.60 64.80 51.60 47.47 53.81 57.00 53.20 54.30 52.90
Kodua_ITMO_task2_1 KoduaITMO2022 74 52.96330674177412 ± 0.00029557613542138784 53.66 51.39 48.84 49.30 41.08 48.93 59.15 51.66 56.48 54.50 63.19 52.88 61.24 51.89 56.25 51.75 55.47 52.13 59.31 52.22 61.08 53.33 55.48 51.47 68.85 56.83 59.08 50.82
Liu_CQUPT_task2_1 LiuCQUPT2022 2 70.16666847945943 ± 0.00036313767268785537 88.45 81.83 70.46 61.14 54.05 55.61 86.04 64.22 68.85 54.45 75.40 64.40 86.31 75.35 81.50 67.12 76.22 64.27 71.40 68.42 89.28 76.04 81.96 70.89 89.98 79.53 94.95 88.25
Liu_CQUPT_task2_2 LiuCQUPT2022 4 69.7042789452307 ± 0.00038117558144412655 88.45 81.83 70.46 61.14 54.05 55.61 86.04 64.22 68.85 54.45 74.73 63.62 85.28 68.74 81.50 67.12 76.22 64.27 71.40 68.42 89.28 76.04 81.96 70.89 90.08 79.80 94.62 88.66
Liu_CQUPT_task2_3 LiuCQUPT2022 3 69.78709634825285 ± 0.0003603881386802032 88.45 81.83 70.46 61.14 54.05 55.61 86.04 64.22 68.85 54.45 78.26 66.39 77.04 73.84 81.50 67.12 76.22 64.27 71.40 68.42 89.28 76.04 81.96 70.89 90.08 79.80 84.14 84.76
Liu_CQUPT_task2_4 LiuCQUPT2022 1 70.97462888079512 ± 0.0003427217026837889 88.45 81.83 70.46 61.14 57.34 57.33 86.04 64.22 68.85 54.45 78.26 66.39 83.87 75.22 81.50 67.12 76.22 64.27 70.11 65.86 89.28 76.04 81.96 70.89 90.08 79.80 93.84 86.80
Siang_NTHU_task2_1 SiangNTHU2022 79 50.59102827288451 ± 0.00027303186490778593 57.84 52.44 44.31 49.56 46.32 50.42 53.15 50.25 38.27 50.91 58.50 51.98 60.33 58.42 78.39 61.24 64.60 55.08 70.73 61.34 82.91 65.22 75.63 65.16 89.87 76.54 88.56 79.06
Tozicka_NSW_task2_1 TozickaNSW2022 38 62.50004793576084 ± 0.00034772455431082163 88.28 74.95 58.44 51.01 42.75 52.14 87.23 72.17 60.35 55.15 62.65 52.16 74.38 70.55 88.24 67.80 72.19 63.80 63.51 60.40 80.22 64.80 70.66 61.00 78.06 62.60 88.73 76.20
Tozicka_NSW_task2_2 TozickaNSW2022 41 62.32235855220123 ± 0.00036151978830945957 90.16 73.71 55.90 50.17 43.55 52.91 83.62 72.21 64.06 55.97 60.47 50.81 74.38 70.55 89.69 71.14 71.31 61.76 63.03 58.87 78.36 54.68 68.73 61.00 81.89 68.44 88.73 76.21
Tozicka_NSW_task2_3 TozickaNSW2022 46 60.461558807688654 ± 0.00035101867648351854 77.78 73.76 52.98 49.95 42.86 51.22 83.61 68.49 60.02 56.72 58.44 51.70 74.38 70.55 86.72 65.40 71.34 60.54 68.24 60.60 72.75 59.56 65.45 53.53 78.44 58.70 88.73 76.20
Tozicka_NSW_task2_4 TozickaNSW2022 43 62.19684397021945 ± 0.00036574848023408654 90.16 73.71 58.44 51.01 42.86 51.22 87.23 72.17 60.35 55.15 60.47 50.81 74.38 70.55 89.69 71.14 72.19 63.81 68.24 60.60 80.22 64.82 70.66 60.95 81.89 68.44 88.73 76.21
Almudevar_UZ_task2_1 AlmudevarUZ2022 50 59.37719756976433 ± 0.0002989255880098096 82.76 68.46 51.58 49.51 47.83 49.35 73.35 52.35 68.07 55.00 63.13 53.36 64.57 54.00 79.14 54.10 56.41 51.29 68.86 56.94 79.44 61.48 71.27 59.34 85.43 68.20 59.95 53.33
Almudevar_UZ_task2_2 AlmudevarUZ2022 45 60.60499062666268 ± 0.00031377932481308624 89.58 77.07 47.34 49.83 47.90 52.47 75.43 56.23 64.84 54.36 68.26 55.78 67.92 55.97 82.62 56.20 60.16 51.18 70.45 60.20 73.88 55.34 63.86 59.15 89.83 77.74 61.77 53.07
Almudevar_UZ_task2_3 AlmudevarUZ2022 47 60.21894525601053 ± 0.00029997437080424505 89.58 77.07 47.10 49.94 48.72 51.62 70.87 52.02 68.07 55.00 65.30 54.61 70.05 55.52 82.62 56.20 60.45 52.08 71.93 60.55 80.99 60.56 71.27 59.34 90.04 77.19 62.64 57.05
Jalalia_AIT_task2_1 JalaliaAIT2022 77 51.338525836051105 ± 0.00029471041821909645 51.35 55.59 44.78 49.55 41.06 50.21 58.20 51.69 60.13 55.52 58.51 53.13 48.68 51.11 63.47 52.75 52.36 50.68 59.38 57.85 63.91 57.90 54.11 51.72 58.28 55.23 48.69 50.48
Zorin_AIRI_task2_1 ZorinAIRI2022 84 43.16915219364365 ± 0.0002620788289881202 37.17 49.65 45.02 49.74 43.23 49.83 40.98 49.49 38.40 50.13 40.83 48.68 39.49 48.92 54.64 52.77 60.43 51.17 54.28 50.10 56.27 50.32 46.42 48.66 63.60 53.43 47.75 51.18
Venkatesh_MERL_task2_1 VenkateshMERL2022 23 65.65983067326367 ± 0.0003545938092026855 93.88 78.67 55.53 54.33 44.50 50.84 86.47 68.54 69.94 61.64 73.64 60.70 78.51 66.08 83.66 65.19 66.33 58.03 71.27 66.13 90.57 71.18 78.26 66.73 90.16 77.60 97.50 92.18
Venkatesh_MERL_task2_2 VenkateshMERL2022 24 65.56948240685708 ± 0.0003696822207797042 93.88 78.67 54.92 54.22 44.29 50.97 82.37 70.76 69.94 61.64 75.96 62.40 77.69 65.39 83.66 65.19 68.31 57.21 71.34 66.16 91.20 75.71 78.26 66.73 90.00 79.55 97.52 92.72
Venkatesh_MERL_task2_3 VenkateshMERL2022 8 67.56529995172603 ± 0.000353158822539437 93.88 78.67 58.23 54.73 48.17 50.34 86.76 79.43 72.54 61.86 73.64 60.70 83.72 62.93 83.66 65.19 66.38 57.27 64.72 61.61 87.47 70.70 73.16 57.59 90.16 77.60 97.35 90.76
Venkatesh_MERL_task2_4 VenkateshMERL2022 10 67.49394662130544 ± 0.0003867088205398765 93.30 75.47 57.30 54.93 46.93 50.33 86.34 78.47 71.96 64.26 75.94 64.29 83.05 64.01 83.68 66.36 66.58 57.84 65.70 62.78 88.73 70.12 78.38 62.40 90.00 77.46 97.09 89.83


Supplementary metrics (recall, precision, and F1 score)

Rank Submission Information Evaluation Dataset
Submission Code Technical
Report
Official
Rank
ToyCar
(F1 score)
ToyCar
(Recall)
ToyCar
(Precision)
ToyTrain
(F1 score)
ToyTrain
(Recall)
ToyTrain
(Precision)
Fan
(F1 score)
Fan
(Recall)
Fan
(Precision)
Gearbox
(F1 score)
Gearbox
(Recall)
Gearbox
(Precision)
Bearing
(F1 score)
Bearing
(Recall)
Bearing
(Precision)
Slider
(F1 score)
Slider
(Recall)
Slider
(Precision)
Valve
(F1 score)
Valve
(Recall)
Valve
(Precision)
DCASE2022_baseline_task2_AE DCASE2022baseline2022 75 51.71 44.04 62.60 0.00 0.00 0.00 0.00 0.00 0.00 67.93 98.27 51.91 58.33 61.81 55.22 55.30 53.97 56.70 29.78 20.49 54.46
DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 68 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 55.14 52.94 57.52 43.31 34.77 57.41 42.52 31.43 65.69 0.00 0.00 0.00
Bai_JLESS_task2_1 BaiJLESS2022 26 85.28 94.80 77.50 0.00 0.00 0.00 25.73 16.50 58.35 66.70 62.01 72.15 64.14 71.76 57.98 71.97 80.66 64.97 61.21 49.20 80.99
Bai_JLESS_task2_2 BaiJLESS2022 44 19.27 12.61 40.86 0.00 0.00 0.00 38.92 32.20 49.17 68.37 64.70 72.48 64.50 67.22 61.98 71.97 80.66 64.97 58.65 45.67 81.94
Bai_JLESS_task2_3 BaiJLESS2022 42 85.28 94.80 77.50 0.00 0.00 0.00 25.73 16.50 58.35 0.00 0.00 0.00 0.00 0.00 0.00 25.16 16.42 53.79 61.21 49.20 80.99
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2022 20 67.62 67.98 67.26 52.40 50.68 54.24 31.77 22.69 52.97 80.93 80.74 81.13 73.37 72.57 74.18 72.58 71.16 74.05 83.96 82.07 85.94
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 5 75.29 74.95 75.63 52.38 49.58 55.52 54.05 49.01 60.26 80.57 80.34 80.81 68.50 68.20 68.80 76.59 76.34 76.84 82.14 80.83 83.49
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2022 52 74.50 73.93 75.09 55.19 55.27 55.12 49.37 43.82 56.54 81.83 81.65 82.00 67.15 66.81 67.50 57.13 56.72 57.55 83.21 82.87 83.55
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2022 12 77.19 76.92 77.46 17.37 10.25 57.02 34.24 26.92 47.03 79.13 78.87 79.40 70.43 69.99 70.88 73.28 72.57 74.02 82.69 81.64 83.76
LEE_KNU_task2_1 LEEKNU2022 82 57.67 65.42 51.56 32.71 26.10 43.79 60.10 72.26 51.44 50.39 50.78 50.00 60.40 79.18 48.82 47.79 46.75 48.88 50.26 49.83 50.70
Narita_AIT_task2_1 NaritaAIT2022 70 66.67 100.00 50.00 66.67 100.00 50.00 66.89 99.66 50.33 66.67 100.00 50.00 66.67 100.00 50.00 66.86 97.61 50.84 64.68 90.86 50.21
Narita_AIT_task2_2 NaritaAIT2022 69 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00 66.67 100.00 50.00
Narita_AIT_task2_3 NaritaAIT2022 76 60.16 65.30 55.77 0.00 0.00 0.00 52.47 48.64 56.96 0.00 0.00 0.00 60.95 71.53 53.10 33.02 25.11 48.19 9.20 5.20 40.00
Narita_AIT_task2_4 NaritaAIT2022 72 66.67 100.00 50.00 66.67 100.00 50.00 52.47 48.64 56.96 0.00 0.00 0.00 66.67 100.00 50.00 33.02 25.11 48.19 64.68 90.86 50.21
Du_NERCSLIP_task2_1 DuNERCSLIP2022 33 64.19 86.32 51.10 66.74 99.32 50.25 63.40 83.11 51.25 77.05 87.94 68.56 69.12 97.30 53.60 50.17 39.31 69.29 0.00 0.00 0.00
Du_NERCSLIP_task2_2 DuNERCSLIP2022 37 64.60 91.19 50.01 66.74 99.32 50.25 63.40 83.11 51.25 77.75 92.31 67.16 69.12 97.30 53.60 31.38 21.35 59.14 0.00 0.00 0.00
Du_NERCSLIP_task2_3 DuNERCSLIP2022 35 64.60 91.19 50.01 66.67 100.00 50.00 41.12 33.74 52.64 77.33 93.36 66.00 69.14 98.99 53.12 31.38 21.35 59.14 0.00 0.00 0.00
Du_NERCSLIP_task2_4 DuNERCSLIP2022 36 64.78 94.21 49.36 66.67 100.00 50.00 59.77 65.73 54.80 73.30 98.57 58.34 69.53 99.66 53.38 14.91 9.01 43.22 36.28 22.41 95.26
Jinhyuk_SNU_task2_1 JinhyukSNU2022 78 41.62 32.89 56.66 66.67 100.00 50.00 0.00 0.00 0.00 69.04 91.91 55.29 62.33 75.94 52.85 47.11 39.15 59.14 21.80 13.76 52.41
Hu_NJU_task2_1 HuNJU2022 65 0.00 0.00 0.00 0.00 0.00 0.00 13.47 7.66 55.52 0.00 0.00 0.00 0.00 0.00 0.00 51.63 45.59 59.51 0.00 0.00 0.00
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 28 67.19 100.00 50.59 67.57 98.18 51.51 56.01 63.98 49.81 66.82 100.00 50.17 67.11 100.00 50.51 68.61 99.31 52.41 74.10 98.95 59.22
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2022 30 66.67 100.00 50.00 67.57 100.00 51.02 58.50 71.37 49.56 66.67 100.00 50.00 66.82 100.00 50.17 67.76 99.31 51.42 70.09 100.00 53.96
Wei_HEU_task2_1 WeiHEU2022 29 26.91 17.55 57.63 45.60 36.81 59.89 0.00 0.00 0.00 62.64 51.12 80.86 36.95 24.98 70.95 21.59 13.45 54.66 49.24 33.19 95.39
Wei_HEU_task2_2 WeiHEU2022 14 85.41 78.80 93.24 50.75 46.23 56.26 36.26 26.45 57.65 78.80 76.60 81.13 61.32 53.04 72.66 67.36 63.28 71.99 70.96 67.71 74.54
Wei_HEU_task2_3 WeiHEU2022 17 83.83 77.33 91.53 50.75 46.23 56.26 25.16 16.23 55.97 79.34 75.92 83.08 36.95 24.98 70.95 66.31 62.21 70.99 73.76 69.71 78.30
Guan_HEU_task2_1 GuanHEU2022 22 0.00 0.00 0.00 59.58 64.75 55.17 48.79 47.55 50.08 73.33 95.65 59.45 67.47 96.70 51.80 0.00 0.00 0.00 76.52 95.87 63.67
Guan_HEU_task2_2 GuanHEU2022 19 73.71 100.00 58.37 0.00 0.00 0.00 48.79 47.55 50.08 75.65 93.16 63.67 68.22 94.30 53.44 0.00 0.00 0.00 0.00 0.00 0.00
Guan_HEU_task2_3 GuanHEU2022 11 74.83 82.49 68.48 53.93 45.71 65.75 54.13 56.84 51.67 68.22 95.93 52.93 15.03 8.41 70.26 54.20 52.72 55.76 77.57 93.60 66.23
Guan_HEU_task2_4 GuanHEU2022 6 78.27 93.46 67.33 55.12 47.75 65.18 54.13 56.84 51.67 72.66 95.58 58.60 16.21 9.34 61.31 64.39 65.53 63.29 78.75 92.22 68.71
Li_CTRI_task2_1 LiCTRI2022 59 69.87 74.12 66.09 39.57 34.10 47.12 0.00 0.00 0.00 54.27 43.07 73.34 67.26 80.16 57.94 29.58 19.25 63.87 0.00 0.00 0.00
Morita_SECOM_task2_1 MoritaSECOM2022 15 69.69 56.47 90.99 0.00 0.00 0.00 15.71 9.09 57.80 42.25 29.92 71.86 0.00 0.00 0.00 67.97 64.01 72.44 78.36 82.04 75.00
Morita_SECOM_task2_2 MoritaSECOM2022 25 87.86 94.76 81.89 0.00 0.00 0.00 0.00 0.00 0.00 68.12 56.18 86.48 0.00 0.00 0.00 61.80 55.09 70.36 69.20 72.10 66.53
Morita_SECOM_task2_3 MoritaSECOM2022 21 69.69 56.47 90.99 0.00 0.00 0.00 0.00 0.00 0.00 68.12 56.18 86.48 0.00 0.00 0.00 61.80 55.09 70.36 76.67 79.12 74.38
Morita_SECOM_task2_4 MoritaSECOM2022 16 69.69 56.47 90.99 0.00 0.00 0.00 0.00 0.00 0.00 68.12 56.18 86.48 0.00 0.00 0.00 61.80 55.09 70.36 0.00 0.00 0.00
Yamashita_GU_task2_1 YamashitaGU2022 53 57.90 49.51 69.72 44.19 34.76 60.63 32.08 24.53 46.34 56.42 49.95 64.82 37.78 27.93 58.38 62.71 53.52 75.71 67.46 55.67 85.60
Yamashita_GU_task2_2 YamashitaGU2022 61 61.39 49.75 80.16 43.87 36.71 54.51 35.18 27.21 49.77 63.93 56.20 74.13 40.10 30.81 57.43 60.61 53.71 69.55 44.54 36.45 57.23
Yamashita_GU_task2_3 YamashitaGU2022 39 61.39 49.75 80.16 44.19 34.76 60.63 35.00 26.11 53.10 63.93 56.20 74.13 57.97 50.23 68.55 60.61 53.71 69.55 67.46 55.67 85.60
Yamashita_GU_task2_4 YamashitaGU2022 48 56.57 45.85 73.84 40.26 30.90 57.73 0.00 0.00 0.00 63.93 56.20 74.13 56.52 48.65 67.45 62.90 55.01 73.41 67.46 55.67 85.60
CHO_SG_task2_1 CHOSG2022 67 69.65 84.99 59.01 66.33 90.90 52.21 63.47 87.45 49.81 66.56 96.20 50.89 0.00 0.00 0.00 62.14 85.54 48.79 65.45 93.44 50.37
CHO_SG_task2_2 CHOSG2022 81 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Li_JAIST_task2 LiJAIST2022 83 66.58 96.96 50.69 0.00 0.00 0.00 0.00 0.00 0.00 14.90 9.71 32.03 64.53 92.72 49.48 0.00 0.00 0.00 30.99 19.20 80.23
Gou_UESTC_task2_1 GouUESTC2022 56 63.10 62.91 63.29 46.35 42.30 51.26 40.05 32.91 51.13 72.48 72.02 72.96 53.96 52.28 55.76 60.61 60.25 60.97 59.78 59.10 60.48
Gou_UESTC_task2_2 GouUESTC2022 58 54.55 54.04 55.08 46.99 43.58 50.97 46.50 41.88 52.26 66.16 66.09 66.22 54.79 54.04 55.57 59.96 59.61 60.31 71.51 70.75 72.29
Gou_UESTC_task2_3 GouUESTC2022 60 67.24 66.60 67.90 48.83 45.22 53.06 45.61 43.84 47.53 65.85 65.13 66.59 63.22 62.93 63.51 60.06 58.40 61.82 51.54 49.72 53.49
Gou_UESTC_task2_4 GouUESTC2022 51 63.10 62.91 63.29 46.35 42.30 51.26 45.61 43.84 47.53 72.48 72.02 72.96 54.16 52.31 56.14 65.07 72.26 59.18 71.51 70.75 72.29
PENG_NJUPT_task2_1 PENGNJUPT2022 57 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 54.22 51.13 57.70 67.33 78.30 59.06 76.81 90.14 66.92 65.46 57.15 76.59
PENG_NJUPT_task2_2 PENGNJUPT2022 62 0.00 0.00 0.00 0.00 0.00 0.00 59.45 62.59 56.61 45.07 36.98 57.69 63.66 68.06 59.79 79.67 92.28 70.09 54.23 49.16 60.45
Nejjar_ETH_task2_1 NejjarETH2022 32 72.84 98.95 57.63 40.59 32.74 53.40 0.00 0.00 0.00 64.47 81.18 53.47 66.74 100.00 50.08 66.67 100.00 50.00 37.11 26.87 60.00
Nejjar_ETH_task2_2 NejjarETH2022 54 72.84 98.95 57.63 40.59 32.74 53.40 0.00 0.00 0.00 64.47 81.18 53.47 66.74 100.00 50.08 66.67 100.00 50.00 37.11 26.87 60.00
Verbitskiy_DS_task2_1 VerbitskiyDS2022 27 66.08 72.32 60.83 56.54 64.15 50.55 21.91 13.70 54.72 78.58 79.34 77.84 60.34 57.99 62.89 65.83 59.64 73.45 83.13 82.92 83.34
Verbitskiy_DS_task2_2 VerbitskiyDS2022 40 66.08 72.32 60.83 46.83 43.54 50.65 21.91 13.70 54.72 72.63 69.00 76.66 60.34 57.99 62.89 61.77 53.13 73.75 83.13 82.92 83.34
Verbitskiy_DS_task2_3 VerbitskiyDS2022 34 72.73 80.20 66.54 55.17 59.16 51.69 27.32 18.27 54.08 72.63 69.00 76.66 60.34 57.99 62.89 61.77 53.13 73.75 83.37 81.88 84.90
Verbitskiy_DS_task2_4 VerbitskiyDS2022 31 73.33 80.77 67.14 54.58 55.20 53.98 27.22 18.31 53.02 72.83 69.38 76.64 60.81 57.70 64.26 61.77 53.13 73.75 83.89 82.54 85.29
Cohen_Technion_task2_1 CohenTechnion2022 64 62.99 60.04 66.25 50.78 50.56 51.00 41.77 36.70 48.47 62.74 62.26 63.22 56.95 53.40 61.01 52.88 49.10 57.31 61.00 60.03 62.00
Cohen_Technion_task2_2 CohenTechnion2022 63 62.28 58.49 66.58 49.46 46.88 52.33 45.49 41.23 50.74 64.84 64.57 65.11 59.83 58.20 61.55 60.24 58.87 61.68 54.21 54.14 54.27
Deng_THU_task2_1 DengTHU2022 9 66.67 100.00 50.00 29.52 19.25 63.29 40.77 37.08 45.27 68.13 97.86 52.26 61.40 61.45 61.34 69.01 98.62 53.07 79.86 99.66 66.62
Deng_THU_task2_2 DengTHU2022 18 66.67 100.00 50.00 48.03 40.70 58.57 45.50 44.71 46.33 68.13 97.86 52.26 60.53 60.09 60.99 68.27 98.95 52.11 78.23 100.00 64.24
Deng_THU_task2_3 DengTHU2022 13 66.67 100.00 50.00 55.17 57.24 53.25 41.69 39.29 44.40 69.25 97.55 53.68 56.72 50.67 64.42 68.93 98.24 53.09 79.86 99.66 66.62
Deng_THU_task2_4 DengTHU2022 7 67.64 100.00 51.11 37.00 28.94 51.28 43.49 38.91 49.30 70.70 95.25 56.21 66.55 98.57 50.24 68.75 98.24 52.88 77.22 94.72 65.17
Liu_BUPT_task2_1 LiuBUPT2022 55 12.24 6.97 50.13 0.00 0.00 0.00 0.00 0.00 0.00 14.99 8.57 59.60 8.52 4.55 66.09 0.00 0.00 0.00 0.00 0.00 0.00
Liu_BUPT_task2_2 LiuBUPT2022 49 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 14.99 8.57 59.60 8.52 4.55 66.09 0.00 0.00 0.00 0.00 0.00 0.00
Kazakova_ITMO_task2_1 KazakovaITMO2022 80 64.79 92.43 49.87 63.01 79.09 52.37 55.94 63.85 49.77 66.70 84.90 54.93 53.50 68.94 43.71 49.85 60.32 42.48 63.49 75.63 54.71
Kazakova_ITMO_task2_2 KazakovaITMO2022 86 27.48 19.11 48.91 28.93 21.75 43.20 25.86 17.13 52.67 37.28 31.66 45.34 12.75 7.85 33.78 25.50 16.76 53.25 0.00 0.00 0.00
Kazakova_ITMO_task2_3 KazakovaITMO2022 85 64.79 92.43 49.87 63.01 79.09 52.37 55.94 63.85 49.77 37.28 31.66 45.34 53.50 68.94 43.71 25.50 16.76 53.25 0.00 0.00 0.00
Kodua_ITMO_task2_1 KoduaITMO2022 74 7.02 3.75 55.05 0.00 0.00 0.00 0.00 0.00 0.00 20.07 11.93 63.05 16.27 9.12 75.43 0.00 0.00 0.00 0.00 0.00 0.00
Liu_CQUPT_task2_1 LiuCQUPT2022 2 77.69 75.86 79.62 64.69 71.15 59.30 41.39 32.52 56.91 79.35 78.82 79.89 65.25 63.81 66.75 70.66 68.49 72.98 76.73 75.45 78.04
Liu_CQUPT_task2_2 LiuCQUPT2022 4 77.69 75.86 79.62 64.69 71.15 59.30 41.39 32.52 56.91 79.35 78.82 79.89 65.25 63.81 66.75 69.59 68.42 70.79 76.82 75.54 78.15
Liu_CQUPT_task2_3 LiuCQUPT2022 3 77.69 75.86 79.62 64.69 71.15 59.30 41.39 32.52 56.91 79.35 78.82 79.89 65.25 63.81 66.75 70.52 68.76 72.38 29.54 19.13 64.85
Liu_CQUPT_task2_4 LiuCQUPT2022 1 77.69 75.86 79.62 64.69 71.15 59.30 45.11 37.68 56.19 79.35 78.82 79.89 65.25 63.81 66.75 70.52 68.76 72.38 71.03 68.47 73.80
Siang_NTHU_task2_1 SiangNTHU2022 79 66.74 100.00 50.08 65.15 95.93 49.33 56.77 67.61 48.93 57.32 59.96 54.90 60.10 77.17 49.22 58.21 65.32 52.50 66.03 93.61 51.00
Tozicka_NSW_task2_1 TozickaNSW2022 38 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Tozicka_NSW_task2_2 TozickaNSW2022 41 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Tozicka_NSW_task2_3 TozickaNSW2022 46 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Tozicka_NSW_task2_4 TozickaNSW2022 43 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Almudevar_UZ_task2_1 AlmudevarUZ2022 50 72.41 68.56 76.72 49.88 48.76 51.06 48.70 44.02 54.49 69.17 68.87 69.47 64.26 60.94 67.97 62.46 59.29 65.99 61.14 60.38 61.91
Almudevar_UZ_task2_2 AlmudevarUZ2022 45 79.18 76.79 81.72 48.67 45.94 51.74 46.85 43.62 50.60 71.63 71.16 72.09 61.79 58.21 65.85 64.93 64.03 65.86 62.56 61.63 63.53
Almudevar_UZ_task2_3 AlmudevarUZ2022 47 79.18 76.79 81.72 48.86 44.39 54.32 49.36 45.19 54.39 66.22 65.52 66.94 64.26 60.94 67.97 64.13 62.93 65.38 61.60 59.98 63.31
Jalalia_AIT_task2_1 JalaliaAIT2022 77 36.45 27.61 53.63 0.00 0.00 0.00 14.09 8.37 44.56 37.97 30.24 51.03 59.71 67.08 53.79 27.09 16.71 71.46 10.14 5.60 53.76
Zorin_AIRI_task2_1 ZorinAIRI2022 84 7.46 3.89 88.89 0.00 0.00 0.00 12.32 7.09 46.88 4.05 2.18 27.91 12.78 7.21 56.46 0.00 0.00 0.00 6.48 3.48 47.06
Venkatesh_MERL_task2_1 VenkateshMERL2022 23 81.19 100.00 68.34 39.09 28.29 63.21 0.00 0.00 0.00 75.69 94.52 63.12 68.43 97.86 52.61 67.73 77.42 60.19 0.00 0.00 0.00
Venkatesh_MERL_task2_2 VenkateshMERL2022 24 78.33 100.00 64.38 42.04 32.52 59.44 59.27 72.85 49.96 71.05 92.11 57.83 67.49 97.86 51.51 70.12 84.80 59.77 74.94 83.65 67.87
Venkatesh_MERL_task2_3 VenkateshMERL2022 8 81.19 100.00 68.34 42.50 31.38 65.84 58.09 69.75 49.78 0.00 0.00 0.00 67.10 96.92 51.31 67.88 77.42 60.43 75.78 82.49 70.08
Venkatesh_MERL_task2_4 VenkateshMERL2022 10 79.58 100.00 66.08 38.29 27.31 64.07 59.51 73.47 50.01 71.33 90.61 58.82 66.87 96.00 51.30 66.22 76.25 58.52 76.31 83.99 69.92


Domain-wise performance

Rank Submission Information Ranking Eveluation Dataset in Source Domain Eveluation Dataset in Target Domain
Submission Code Technical
Report
Official
Rank
Official
Score
Harmonic mean
(AUC, source)
ToyCar
(AUC, source)
ToyCar
(pAUC, source)
ToyTrain
(AUC, source)
ToyTrain
(pAUC, source)
Fan
(AUC, source)
Fan
(pAUC, source)
Gearbox
(AUC, source)
Gearbox
(pAUC, source)
Bearing
(AUC, source)
Bearing
(pAUC, source)
Slider
(AUC, source)
Slider
(pAUC, source)
Valve
(AUC, source)
Valve
(pAUC, source)
Harmonic mean
(AUC, target)
ToyCar
(AUC, target)
ToyCar
(pAUC, target)
ToyTrain
(AUC, target)
ToyTrain
(pAUC, target)
Fan
(AUC, target)
Fan
(pAUC, target)
Gearbox
(AUC, target)
Gearbox
(pAUC, target)
Bearing
(AUC, target)
Bearing
(pAUC, target)
Slider
(AUC, target)
Slider
(pAUC, target)
Valve
(AUC, target)
Valve
(pAUC, target)
DCASE2022_baseline_task2_AE DCASE2022baseline2022 75 52.942 64.23 82.04 71.00 50.18 49.68 84.56 52.21 60.18 54.32 63.38 47.37 75.66 54.00 56.96 54.32 45.13 58.22 58.43 41.96 49.30 37.33 49.72 62.28 51.50 59.28 55.49 56.46 54.19 53.75 50.74
DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 68 54.017 58.48 35.76 48.58 40.98 48.89 79.78 65.95 78.92 48.84 57.22 50.74 94.60 52.21 72.28 69.79 50.49 44.54 54.53 53.92 51.42 46.88 53.48 47.99 48.43 58.43 52.45 58.44 53.25 72.87 64.31
Bai_JLESS_task2_1 BaiJLESS2022 26 63.949 74.16 86.32 68.84 42.48 49.26 80.40 52.32 62.42 50.16 65.04 51.63 96.10 57.95 92.60 65.16 59.35 93.43 77.92 54.15 51.63 42.91 50.43 74.03 66.28 63.82 56.71 74.40 66.87 81.49 69.49
Bai_JLESS_task2_2 BaiJLESS2022 44 61.102 68.53 62.56 60.11 56.89 51.42 80.08 62.74 67.12 51.89 67.54 51.42 96.10 57.95 68.66 60.32 57.23 51.87 52.11 49.65 52.12 49.46 54.53 74.11 70.53 64.43 56.56 74.40 66.87 80.38 68.44
Bai_JLESS_task2_3 BaiJLESS2022 42 62.208 73.27 86.32 68.84 42.48 49.26 80.40 52.32 68.52 47.47 57.74 48.63 93.20 61.79 92.60 65.16 58.69 93.43 77.92 54.15 51.63 42.91 50.43 71.69 51.29 70.79 56.91 63.04 54.31 81.49 69.49
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2022 20 66.503 81.35 65.68 53.79 52.20 49.53 96.42 57.11 93.24 47.58 85.30 55.37 99.80 59.53 98.10 93.37 59.88 65.34 76.55 57.60 52.43 44.55 51.85 80.30 69.86 77.34 62.96 77.83 58.68 91.43 82.22
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 5 68.223 79.36 75.46 56.11 54.18 50.84 87.90 60.16 94.74 47.37 77.96 48.11 90.54 52.11 95.78 92.26 66.21 83.04 76.55 55.19 51.98 53.46 51.79 82.72 70.37 71.27 60.36 78.90 59.45 90.11 81.63
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2022 52 59.034 50.86 72.36 54.26 54.04 54.95 91.54 56.79 95.14 49.21 68.78 48.00 6.22 47.37 99.54 98.21 66.47 77.17 76.82 55.67 52.36 54.65 54.42 86.61 74.43 74.79 58.64 79.49 59.31 89.12 82.14
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2022 12 67.142 82.00 72.04 56.00 54.96 49.47 89.72 59.95 89.78 47.53 71.84 51.53 88.18 51.53 99.04 96.58 60.88 88.96 78.69 44.34 51.83 49.80 55.23 82.54 67.39 77.94 58.10 80.10 61.81 91.25 86.36
LEE_KNU_task2_1 LEEKNU2022 82 49.267 49.88 58.05 50.35 43.80 50.28 59.96 49.49 43.56 48.99 58.56 50.40 49.61 49.39 46.26 48.98 47.71 52.49 51.93 45.32 49.70 48.07 50.43 51.15 49.60 45.84 49.83 47.06 49.25 49.98 52.38
Narita_AIT_task2_1 NaritaAIT2022 70 53.512 55.48 84.74 62.89 60.00 53.37 97.80 65.68 84.46 71.74 69.92 49.42 99.66 56.16 100.00 93.95 51.87 55.61 51.89 45.86 50.62 55.24 53.18 52.76 53.23 53.21 49.83 44.11 50.13 46.77 54.21
Narita_AIT_task2_2 NaritaAIT2022 69 53.850 56.58 84.04 73.63 61.46 52.21 97.44 64.84 73.36 62.47 69.30 50.95 99.76 58.74 90.74 66.47 52.20 57.82 54.07 49.35 50.04 53.91 52.88 50.61 52.60 54.30 50.42 45.14 50.57 48.22 51.46
Narita_AIT_task2_3 NaritaAIT2022 76 52.359 54.77 84.44 56.63 58.12 52.42 90.36 56.63 65.14 57.95 71.58 54.11 78.64 55.63 85.90 74.47 50.45 54.19 51.41 47.93 50.16 50.54 52.31 48.68 50.95 53.42 50.49 47.27 50.10 46.18 52.01
Narita_AIT_task2_4 NaritaAIT2022 72 53.274 55.57 84.04 73.63 61.46 52.21 90.36 56.63 65.14 57.95 69.30 50.95 78.64 55.63 100.00 93.95 51.46 57.82 54.07 49.35 50.04 50.54 52.31 48.68 50.95 54.30 50.42 47.27 50.10 46.77 54.21
Du_NERCSLIP_task2_1 DuNERCSLIP2022 33 63.303 76.37 64.12 49.58 55.92 48.32 74.18 50.32 96.30 51.00 60.56 49.47 96.70 50.84 94.42 92.84 59.07 63.75 61.45 58.60 52.41 50.14 51.86 80.18 62.57 68.60 55.28 63.50 57.78 85.85 73.17
Du_NERCSLIP_task2_2 DuNERCSLIP2022 37 62.580 75.77 53.54 48.89 56.46 49.21 74.18 50.32 96.52 50.42 60.56 49.47 97.90 51.00 94.40 92.84 57.61 56.62 59.97 59.65 53.01 50.14 51.86 79.27 63.38 68.60 55.28 63.67 55.82 86.57 75.73
Du_NERCSLIP_task2_3 DuNERCSLIP2022 35 62.787 76.82 53.54 48.89 56.46 48.89 81.10 50.05 96.88 50.47 64.18 49.37 97.90 51.00 94.40 92.84 57.59 56.62 59.97 59.97 53.81 50.33 51.78 79.58 62.71 69.01 54.79 63.67 55.82 86.57 75.73
Du_NERCSLIP_task2_4 DuNERCSLIP2022 36 62.749 75.12 48.20 49.11 51.82 50.95 85.74 50.58 96.40 49.53 66.06 49.47 100.00 53.74 94.58 92.95 58.47 52.15 59.35 60.07 54.25 53.05 53.15 78.23 60.84 68.73 52.44 67.79 56.28 87.45 77.96
Jinhyuk_SNU_task2_1 JinhyukSNU2022 78 51.027 62.46 78.94 59.42 49.94 49.11 83.32 55.89 54.60 51.68 63.42 47.47 72.58 52.63 49.56 51.89 42.54 48.24 53.02 41.54 49.02 37.54 49.99 59.26 51.19 58.75 55.24 56.33 53.87 49.17 50.79
Hu_NJU_task2_1 HuNJU2022 65 55.075 61.16 73.22 51.32 42.83 49.47 61.70 52.47 62.46 53.47 66.43 51.74 62.44 52.00 55.78 50.84 51.90 68.57 57.76 52.03 51.30 47.19 50.92 63.95 54.54 61.74 57.88 59.17 52.78 44.99 49.43
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 28 63.724 85.37 97.18 68.42 77.68 52.21 77.30 49.05 96.24 58.37 82.56 61.53 95.26 52.53 98.36 90.47 54.12 76.92 65.05 58.65 57.02 36.85 48.84 80.48 58.72 74.70 66.23 63.35 55.67 82.23 67.37
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2022 30 63.557 85.03 97.26 70.53 81.82 52.47 72.44 50.21 92.74 60.84 80.54 60.26 93.40 52.16 98.84 89.11 54.13 75.08 64.51 58.04 57.48 38.59 48.55 78.47 58.27 73.72 65.33 64.15 55.33 79.64 65.87
Wei_HEU_task2_1 WeiHEU2022 29 63.603 77.89 67.98 54.79 67.75 54.47 76.27 50.26 93.58 51.11 71.10 52.74 95.84 56.00 91.68 92.42 58.46 73.79 62.78 62.06 53.53 41.57 51.93 76.70 58.63 66.97 55.09 68.85 55.47 84.64 77.94
Wei_HEU_task2_2 WeiHEU2022 14 67.122 73.31 97.22 92.16 62.38 48.95 93.66 52.47 93.66 66.42 72.40 49.00 76.40 52.47 93.74 79.58 67.96 97.16 87.40 56.13 54.05 45.19 51.58 83.66 59.84 70.09 60.63 74.01 63.04 82.83 65.07
Wei_HEU_task2_3 WeiHEU2022 17 66.652 76.76 96.16 90.26 62.38 48.95 93.82 52.58 94.66 61.05 71.10 52.74 82.04 52.89 94.90 85.95 64.51 95.82 86.69 56.13 54.05 44.77 49.59 83.59 62.43 66.97 55.09 73.25 62.35 84.73 67.84
Guan_HEU_task2_1 GuanHEU2022 22 66.395 77.50 95.16 84.95 64.80 48.68 98.14 56.05 95.86 72.79 77.88 49.89 84.10 53.21 94.28 77.84 63.51 87.99 77.55 57.01 54.07 44.74 50.95 81.44 61.36 70.58 61.17 72.61 62.01 80.23 62.49
Guan_HEU_task2_2 GuanHEU2022 19 66.532 75.96 94.82 90.21 64.72 48.74 98.14 56.05 94.92 67.47 71.70 48.89 79.94 51.84 93.64 74.79 65.15 93.99 84.84 57.03 54.08 44.74 50.95 81.99 59.74 71.47 60.09 72.43 62.29 80.16 60.91
Guan_HEU_task2_3 GuanHEU2022 11 67.419 81.36 88.12 63.05 70.70 49.74 98.00 56.21 95.68 58.63 69.24 50.00 93.86 56.26 97.00 91.79 63.89 84.46 72.73 60.31 56.32 50.92 53.25 79.04 58.34 66.81 53.64 72.31 62.40 86.65 76.48
Guan_HEU_task2_4 GuanHEU2022 6 68.037 81.43 92.62 77.16 70.24 49.63 98.00 56.21 94.92 56.00 68.54 49.74 91.24 52.79 97.20 92.58 64.88 90.56 78.67 60.18 56.41 50.92 53.25 79.32 58.52 66.98 53.35 72.59 62.49 87.87 77.01
Li_CTRI_task2_1 LiCTRI2022 59 58.128 66.11 88.76 64.53 50.86 49.53 81.94 52.84 73.54 48.74 56.36 47.79 94.64 55.89 71.32 68.95 55.06 73.69 58.38 48.36 50.01 36.43 49.65 68.24 54.17 64.90 51.70 61.11 50.98 84.17 74.01
Morita_SECOM_task2_1 MoritaSECOM2022 15 66.825 77.14 78.18 65.95 58.38 49.74 86.84 57.53 82.70 56.21 75.26 57.00 99.26 49.42 98.94 93.42 63.80 96.04 76.34 58.29 55.91 46.37 51.62 79.37 60.46 73.82 64.58 64.55 65.62 82.62 66.03
Morita_SECOM_task2_2 MoritaSECOM2022 25 65.088 80.14 89.42 75.68 60.98 50.84 87.16 52.11 90.70 67.95 75.88 50.00 98.84 53.05 97.74 89.53 58.23 94.62 77.94 61.36 53.87 41.79 50.00 88.43 73.64 65.73 54.17 62.09 64.44 69.49 63.39
Morita_SECOM_task2_3 MoritaSECOM2022 21 66.399 80.90 78.18 65.95 60.98 50.84 87.16 52.11 90.70 67.95 74.82 51.53 98.84 53.05 98.20 92.16 59.96 96.04 76.34 61.36 53.87 41.79 50.00 88.43 73.64 71.63 62.15 62.09 64.44 77.68 65.88
Morita_SECOM_task2_4 MoritaSECOM2022 16 66.724 78.14 78.18 65.95 60.98 50.84 88.40 56.16 90.70 67.95 74.82 51.53 98.84 53.05 96.86 86.68 62.35 96.04 76.34 61.36 53.87 42.70 50.83 88.43 73.64 71.63 62.15 62.09 64.44 79.72 64.57
Yamashita_GU_task2_1 YamashitaGU2022 53 58.941 69.82 83.64 75.11 65.86 47.79 72.36 52.16 59.82 62.95 57.86 47.37 78.50 53.00 95.52 84.26 53.58 60.46 57.33 54.01 49.47 39.02 49.13 69.41 54.65 55.72 50.88 71.26 61.68 84.44 71.18
Yamashita_GU_task2_2 YamashitaGU2022 61 57.965 69.08 92.04 81.00 66.56 49.42 81.56 54.16 74.82 63.84 64.68 47.37 76.84 53.47 58.04 54.95 52.44 74.41 66.35 50.92 49.75 39.85 49.64 75.78 59.03 55.76 51.92 68.58 58.77 56.95 51.31
Yamashita_GU_task2_3 YamashitaGU2022 39 62.387 73.21 92.04 81.00 65.86 47.79 83.10 51.89 74.82 63.84 64.54 47.47 76.84 53.47 95.52 84.26 58.70 74.41 66.35 54.01 49.47 42.92 49.49 75.78 59.03 62.67 53.67 68.58 58.77 84.44 71.18
Yamashita_GU_task2_4 YamashitaGU2022 48 60.209 73.22 89.46 78.74 66.18 47.89 83.00 51.84 74.82 63.84 64.06 47.37 78.88 53.47 95.52 84.26 53.57 65.30 60.22 53.72 49.56 35.25 49.30 75.78 59.03 61.83 53.36 70.68 61.76 84.44 71.18
CHO_SG_task2_1 CHOSG2022 67 54.469 54.69 65.78 57.95 51.42 51.26 57.86 50.26 62.26 50.63 54.34 54.84 46.18 49.68 34.82 49.32 55.93 73.73 56.30 55.46 51.70 48.17 50.89 63.84 56.48 52.32 50.61 48.02 50.96 60.29 55.57
CHO_SG_task2_2 CHOSG2022 81 49.288 47.39 36.10 49.53 46.46 49.95 38.80 49.16 54.80 51.89 45.10 48.53 54.34 52.37 47.66 49.47 49.91 53.26 51.42 50.04 50.64 48.20 51.32 49.00 51.97 51.42 50.66 43.95 49.90 50.75 49.82
Li_JAIST_task2 LiJAIST2022 83 48.939 47.60 69.62 57.16 50.66 52.26 69.06 49.63 45.36 47.63 63.54 51.89 28.64 47.37 91.76 89.21 46.65 56.81 56.35 47.23 50.12 42.20 50.18 44.19 48.71 41.04 50.03 33.28 48.10 81.19 73.56
Gou_UESTC_task2_1 GouUESTC2022 56 58.723 64.02 59.78 53.84 56.22 53.05 79.76 53.42 79.74 58.42 75.88 51.11 93.48 58.79 64.92 55.47 57.11 71.05 68.33 50.33 50.02 48.22 51.06 78.30 67.54 52.77 50.78 60.60 57.83 60.73 51.92
Gou_UESTC_task2_2 GouUESTC2022 58 58.434 61.07 50.28 48.74 51.40 49.32 70.84 52.47 71.10 62.32 73.56 55.68 86.42 59.79 79.64 68.16 57.98 64.23 60.85 47.56 49.96 46.33 52.74 71.46 62.26 54.72 51.20 57.37 55.02 81.41 68.49
Gou_UESTC_task2_3 GouUESTC2022 60 58.093 61.64 74.34 48.63 46.98 49.53 76.68 51.89 65.44 63.95 73.64 51.68 87.96 59.21 50.40 48.68 56.55 70.23 66.90 50.94 51.57 39.79 48.94 76.81 67.51 64.47 56.57 63.64 58.45 56.83 55.23
Gou_UESTC_task2_4 GouUESTC2022 51 59.169 65.32 59.78 53.84 56.22 53.05 76.68 51.89 79.74 58.42 44.14 48.47 76.72 52.89 79.64 68.16 56.06 71.05 68.33 50.33 50.02 39.79 48.94 78.30 67.54 61.19 53.14 58.57 53.83 81.41 68.49
PENG_NJUPT_task2_1 PENGNJUPT2022 57 58.473 66.63 49.90 51.42 41.24 48.00 77.20 52.63 85.36 51.63 72.08 48.95 94.44 57.68 85.62 86.21 54.47 59.04 58.73 50.29 51.68 47.80 50.87 55.18 51.03 60.85 55.92 72.58 58.68 76.67 68.80
PENG_NJUPT_task2_2 PENGNJUPT2022 62 57.379 64.33 47.50 48.21 39.04 48.26 80.66 52.00 89.26 58.05 62.70 50.42 98.50 56.32 92.66 91.42 53.53 54.55 58.16 55.53 52.31 46.50 50.90 52.83 49.96 65.74 52.44 73.46 61.98 59.86 65.62
Nejjar_ETH_task2_1 NejjarETH2022 32 63.346 76.52 91.24 78.37 51.64 51.21 87.80 55.89 73.74 57.37 74.62 57.68 98.50 54.32 76.64 57.79 58.08 89.31 79.95 64.20 54.87 44.92 49.84 71.32 60.03 67.91 58.82 55.80 52.42 76.21 63.47
Nejjar_ETH_task2_2 NejjarETH2022 54 58.850 60.87 91.24 78.37 51.64 51.21 12.20 48.89 73.74 57.37 74.62 57.68 98.50 54.32 76.64 57.79 57.63 89.31 79.95 64.20 54.87 43.80 48.60 71.32 60.03 67.91 58.82 55.80 52.42 76.21 63.47
Verbitskiy_DS_task2_1 VerbitskiyDS2022 27 63.831 74.27 72.54 53.63 48.32 47.89 88.80 54.05 85.18 67.89 69.80 60.16 95.26 55.11 98.58 94.42 58.57 71.44 68.38 53.01 51.68 40.77 50.37 86.71 76.45 64.63 56.49 69.23 57.79 91.49 78.54
Verbitskiy_DS_task2_2 VerbitskiyDS2022 40 62.382 75.58 72.54 53.63 41.62 50.05 88.80 54.05 89.32 63.58 69.80 60.16 94.18 54.47 98.58 94.42 55.63 71.44 68.38 54.66 51.18 40.77 50.37 77.04 63.37 64.63 56.49 63.34 55.80 91.49 78.54
Verbitskiy_DS_task2_3 VerbitskiyDS2022 34 62.903 74.41 70.38 48.63 46.94 48.63 80.30 52.95 89.32 63.58 69.80 60.16 94.18 54.47 98.22 94.21 57.24 78.39 69.04 54.97 53.51 39.96 50.07 77.04 63.37 64.63 56.49 63.34 55.80 93.05 82.18
Verbitskiy_DS_task2_4 VerbitskiyDS2022 31 63.393 75.74 72.32 48.63 42.28 49.74 80.78 53.32 89.32 63.26 70.54 61.68 94.18 54.47 98.08 94.11 57.49 79.40 69.67 57.84 52.78 39.89 50.11 77.03 63.27 67.41 56.99 63.34 55.80 93.00 84.19
Cohen_Technion_task2_1 CohenTechnion2022 64 55.640 69.03 81.38 60.21 50.48 48.26 67.22 49.63 78.92 52.53 74.37 49.52 84.96 49.37 74.00 56.95 49.05 70.10 58.63 53.12 51.74 41.67 49.53 60.03 49.82 55.82 57.37 50.75 49.33 64.82 53.45
Cohen_Technion_task2_2 CohenTechnion2022 63 56.763 66.63 89.24 68.37 45.92 49.63 80.92 52.84 67.60 52.21 64.02 53.95 86.58 55.84 58.90 55.11 51.69 73.37 64.92 51.53 50.10 39.78 49.64 67.29 51.79 64.48 59.98 60.59 53.11 54.38 51.13
Deng_THU_task2_1 DengTHU2022 9 67.527 75.30 77.08 57.84 59.64 52.95 79.24 64.11 98.18 52.42 66.62 53.21 96.56 51.32 97.66 92.95 64.63 80.03 69.77 63.60 56.06 40.33 51.61 85.96 74.89 66.52 59.29 79.31 70.61 92.22 85.08
Deng_THU_task2_2 DengTHU2022 18 66.533 77.65 77.62 58.42 56.98 52.74 84.02 59.58 98.18 52.42 72.28 54.68 97.28 51.42 98.84 92.74 60.74 80.69 70.73 63.03 55.68 36.97 50.92 85.96 74.89 65.25 59.73 78.24 68.43 92.23 85.15
Deng_THU_task2_3 DengTHU2022 13 67.134 74.45 76.22 57.26 53.26 52.95 77.10 61.63 98.34 52.47 67.52 53.53 96.28 51.32 97.66 92.95 64.28 78.78 69.46 60.92 55.14 40.41 51.50 87.32 76.61 66.91 59.29 79.62 70.74 92.22 85.08
Deng_THU_task2_4 DengTHU2022 7 67.624 77.16 81.74 59.00 47.74 51.21 86.20 51.47 95.14 57.32 72.20 52.00 95.96 51.26 96.86 92.58 64.52 89.96 72.81 63.25 54.23 40.56 49.73 83.61 68.59 71.12 62.32 79.17 71.43 88.37 82.59
Liu_BUPT_task2_1 LiuBUPT2022 55 58.807 72.12 50.12 51.21 46.58 48.42 78.46 53.37 92.94 48.11 83.68 52.53 98.44 58.42 90.92 92.32 51.98 57.89 52.46 45.58 50.07 42.17 49.55 66.36 51.65 67.75 58.33 67.64 60.11 84.07 78.48
Liu_BUPT_task2_2 LiuBUPT2022 49 59.768 74.74 81.20 68.37 52.68 49.63 77.84 53.05 92.94 48.11 83.68 52.53 98.44 58.42 90.92 92.32 51.93 50.43 56.06 55.91 51.61 39.61 49.73 66.36 51.65 67.75 58.33 67.64 60.11 84.07 78.48
Kazakova_ITMO_task2_1 KazakovaITMO2022 80 49.896 47.71 61.86 54.79 57.76 50.95 35.72 48.89 60.82 50.00 43.08 51.95 62.30 51.58 65.88 55.74 50.64 57.69 54.94 50.12 50.63 52.66 50.79 51.33 49.47 38.92 49.50 37.07 49.64 62.23 55.73
Kazakova_ITMO_task2_2 KazakovaITMO2022 86 37.563 32.78 55.58 48.21 59.02 52.47 24.84 48.58 40.30 47.68 46.78 52.47 26.02 49.21 7.18 47.58 33.92 28.82 47.56 48.31 50.34 50.96 50.47 56.50 52.28 48.82 52.28 53.84 53.03 16.57 47.97
Kazakova_ITMO_task2_3 KazakovaITMO2022 85 39.579 33.94 61.86 54.79 57.76 50.95 35.72 48.89 40.30 47.68 43.08 51.95 26.02 49.21 7.18 47.58 37.40 57.69 54.94 50.12 50.63 52.66 50.79 56.50 52.28 38.92 49.50 53.84 53.03 16.57 47.97
Kodua_ITMO_task2_1 KoduaITMO2022 74 52.963 61.04 59.42 52.00 48.78 48.00 74.18 49.47 71.56 54.63 71.68 62.32 85.06 48.84 67.00 53.16 48.03 52.64 51.28 48.85 49.57 37.72 48.82 57.17 51.10 54.18 53.17 60.10 53.77 60.20 51.65
Liu_CQUPT_task2_1 LiuCQUPT2022 2 70.167 81.06 88.18 80.05 61.76 57.95 90.00 65.05 96.68 68.26 71.90 54.21 95.36 68.21 94.38 82.26 67.55 88.51 82.19 72.50 61.82 50.05 54.04 84.19 63.47 68.27 54.50 72.37 63.69 84.85 74.11
Liu_CQUPT_task2_2 LiuCQUPT2022 4 69.704 80.71 88.18 80.05 61.76 57.95 90.00 65.05 96.68 68.26 71.90 54.21 96.92 66.47 95.40 75.74 67.45 88.51 82.19 72.50 61.82 50.05 54.04 84.19 63.47 68.27 54.50 71.45 63.07 83.51 67.50
Liu_CQUPT_task2_3 LiuCQUPT2022 3 69.787 79.83 88.18 80.05 61.76 57.95 90.00 65.05 96.68 68.26 71.90 54.21 95.16 68.37 87.60 80.05 67.22 88.51 82.19 72.50 61.82 50.05 54.04 84.19 63.47 68.27 54.50 75.57 66.00 75.22 72.71
Liu_CQUPT_task2_4 LiuCQUPT2022 1 70.975 77.34 88.18 80.05 61.76 57.95 74.35 72.26 96.68 68.26 71.90 54.21 95.16 68.37 93.80 81.68 72.12 88.51 82.19 72.50 61.82 54.84 55.06 84.19 63.47 68.27 54.50 75.57 66.00 82.13 74.04
Siang_NTHU_task2_1 SiangNTHU2022 79 50.591 51.47 63.22 55.79 50.38 50.84 49.12 52.21 57.00 49.21 49.62 51.84 52.56 50.61 81.94 62.95 48.58 56.88 51.82 43.27 49.31 45.80 50.08 52.45 50.46 36.59 50.73 59.85 52.26 57.31 57.59
Tozicka_NSW_task2_1 TozickaNSW2022 38 62.500 71.02 90.20 81.74 61.58 50.89 86.34 62.42 91.82 81.47 59.08 50.92 83.04 49.66 86.72 86.74 58.34 87.90 73.72 57.85 51.03 38.83 50.48 86.36 70.56 60.61 56.09 59.71 52.69 72.33 68.01
Tozicka_NSW_task2_2 TozickaNSW2022 41 62.322 70.96 90.76 79.84 60.22 50.21 89.86 65.16 84.96 73.89 58.86 51.56 83.82 48.42 86.72 86.74 58.17 90.04 72.60 55.10 50.16 39.48 50.99 83.36 71.89 65.22 56.95 57.28 51.32 72.33 68.01
Tozicka_NSW_task2_3 TozickaNSW2022 46 60.462 69.73 89.28 77.37 56.54 49.79 88.06 56.26 84.94 78.89 60.06 52.92 81.74 49.32 86.72 86.74 54.65 75.83 73.08 52.32 49.98 38.87 50.32 83.35 66.73 60.01 57.55 55.29 52.21 72.33 68.01
Tozicka_NSW_task2_4 TozickaNSW2022 43 62.197 72.55 90.76 79.84 61.58 50.89 88.06 56.26 91.82 81.47 59.08 50.92 83.82 48.42 86.72 86.74 57.09 90.04 72.60 57.85 51.03 38.87 50.32 86.36 70.56 60.61 56.09 57.28 51.32 72.33 68.01
Almudevar_UZ_task2_1 AlmudevarUZ2022 50 59.377 71.63 92.68 70.47 53.64 47.63 89.36 50.32 83.72 47.89 78.62 55.63 94.92 48.05 73.62 61.05 55.41 81.02 68.07 51.19 49.90 43.76 49.16 71.58 53.34 66.29 54.87 59.17 54.56 63.02 52.78
Almudevar_UZ_task2_2 AlmudevarUZ2022 45 60.605 72.09 93.90 79.68 52.52 49.74 91.00 60.89 87.82 49.05 73.42 51.21 95.06 49.79 79.98 62.74 55.86 88.76 76.56 46.42 49.85 43.76 51.06 73.36 57.92 63.36 55.04 64.62 57.16 65.93 54.79
Almudevar_UZ_task2_3 AlmudevarUZ2022 47 60.219 73.43 93.90 79.68 52.84 50.47 90.36 58.74 82.22 47.58 78.62 55.63 94.46 49.05 68.60 55.63 54.99 88.76 76.56 46.10 49.84 44.61 50.40 68.96 53.00 66.29 54.87 61.50 55.88 70.34 55.49
Jalalia_AIT_task2_1 JalaliaAIT2022 77 51.339 61.03 79.60 67.00 48.52 49.68 84.08 52.89 59.10 54.47 62.92 47.74 75.94 52.53 52.02 52.89 43.61 47.95 53.76 44.09 49.52 37.25 49.70 58.03 51.17 59.60 57.39 55.95 53.26 48.06 50.76
Zorin_AIRI_task2_1 ZorinAIRI2022 84 43.169 36.40 34.54 47.95 50.80 49.79 20.68 48.47 25.56 47.58 33.32 50.00 45.60 48.68 42.58 49.68 45.84 37.74 50.01 44.02 49.73 55.28 50.12 46.60 49.89 39.61 50.16 40.00 48.68 38.93 48.77
Venkatesh_MERL_task2_1 VenkateshMERL2022 23 65.660 78.99 87.40 74.21 63.48 50.63 72.96 52.42 88.16 55.74 67.50 52.58 99.88 56.58 98.08 73.95 59.32 95.30 79.63 54.17 55.13 41.28 50.53 86.13 71.84 70.45 63.84 69.96 61.60 75.50 64.70
Venkatesh_MERL_task2_2 VenkateshMERL2022 24 65.569 78.79 87.40 74.21 62.34 49.95 74.08 52.58 77.28 67.47 67.50 52.58 99.88 54.00 97.40 71.47 58.83 95.30 79.63 53.65 55.16 40.99 50.66 83.48 71.45 70.45 63.84 72.48 64.41 74.66 64.29
Venkatesh_MERL_task2_3 VenkateshMERL2022 8 67.565 80.30 87.40 74.21 66.68 51.26 73.38 51.32 80.94 68.74 75.50 48.95 99.88 56.58 94.90 59.95 62.69 95.30 79.63 56.79 55.48 45.07 50.15 88.02 81.98 71.97 65.30 69.96 61.60 81.80 63.56
Venkatesh_MERL_task2_4 VenkateshMERL2022 10 67.494 79.90 83.04 64.21 66.46 50.84 71.98 51.68 82.52 67.05 75.10 52.58 99.88 62.53 93.60 64.26 62.10 95.67 78.22 55.76 55.83 43.88 50.07 87.15 81.24 71.36 67.24 72.46 64.66 81.22 63.96



System characteristics

Summary of the submitted system characteristics.

Rank Submission
Code
Technical
Report
Classifier System
Complexity
Acoustic
Feature
Data
Augmentation
Decision Making System
Embeddings
Subsystem
Conut
External
Data Usage
Front End
System
75 DCASE2022_baseline_task2_AE DCASE2022baseline2022 AE 269992 log-mel energies
68 DCASE2022_baseline_task2_MNV2 DCASE2022baseline2022 MNV2 269992 log-mel energies
26 Bai_JLESS_task2_1 BaiJLESS2022 CNN, Transformer, ensemble, LOF 567110 spectrogram, log-mel energies fmix, mixup vote 3
44 Bai_JLESS_task2_2 BaiJLESS2022 CNN, ensemble, LOF 995590 log-mel energies fmix, mixup vote 3
42 Bai_JLESS_task2_3 BaiJLESS2022 CNN, ensemble, LOF 567110 spectrogram fmix, mixup vote 3
20 Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2022 GMM,LOF,KNN,CNN,Transformer,Conformer 48789168 spectrogram mixup,gaussian noise average PyTorch Image Models 20 pre-trained model,Audioseet
5 Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2022 GMM,LOF,KNN,CNN,Transformer,Conformer 48789168 spectrogram mixup,gaussian noise average PyTorch Image Models 20 pre-trained model,Audioseet
52 Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2022 GMM,LOF,KNN,CNN,Transformer,Conformer 48789168 spectrogram mixup,gaussian noise average PyTorch Image Models 20 pre-trained model,Audioseet
12 Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2022 GMM,LOF,KNN,CNN,Transformer,Conformer 48789168 spectrogram mixup,gaussian noise average PyTorch Image Models 20 pre-trained model,Audioseet
82 LEE_KNU_task2_1 LEEKNU2022 Contrastive learning, k-NN 11496000 log-mel energies Harmonics modification, Temporal masking, F0 masking,
70 Narita_AIT_task2_1 NaritaAIT2022 EfficientNet-B1, Mahalanobis 7794184 log-mel energies Mixup, Time Masking, Frequency Masking PyTorch Image Models (EfficientNet-B1) pre-trained model
69 Narita_AIT_task2_2 NaritaAIT2022 EfficientNet-B1, Mahalanobis 7794184 log-mel energies Mixup, Frequency Masking PyTorch Image Models (EfficientNet-B1) pre-trained model
76 Narita_AIT_task2_3 NaritaAIT2022 EfficientNet-B1, Mahalanobis 7794184 log-mel energies Mixup, Time Masking, Frequency Masking, SevenBandParametricEQ PyTorch Image Models (EfficientNet-B1) pre-trained model
72 Narita_AIT_task2_4 NaritaAIT2022 EfficientNet-B1, Mahalanobis 23382552 log-mel energies Mixup, Time Masking, Frequency Masking, SevenBandParametricEQ PyTorch Image Models (EfficientNet-B1) 3 pre-trained model
33 Du_NERCSLIP_task2_1 DuNERCSLIP2022 AE, Section ID classification, binary classification, cosine distance, ensemble 182M log-mel energies, STFT, raw waveform mixup, TFmask, volume perturbation weighted average SSAST traind on AudioSet data 4 simulation of anomalous samples, pre-trained model
37 Du_NERCSLIP_task2_2 DuNERCSLIP2022 AE, Section ID classification, binary classification, cosine distance, ensemble 183M log-mel energies, STFT, raw waveform mixup, TFmask, volume perturbation weighted average SSAST traind on AudioSet data 4 simulation of anomalous samples, pre-trained model
35 Du_NERCSLIP_task2_3 DuNERCSLIP2022 AE, Section ID classification, binary classification, cosine distance, ensemble 184M log-mel energies, STFT, raw waveform mixup, TFmask, volume perturbation weighted average SSAST traind on AudioSet data 4 simulation of anomalous samples, pre-trained model
36 Du_NERCSLIP_task2_4 DuNERCSLIP2022 AE, Section ID classification, binary classification, cosine distance, ensemble 185M log-mel energies, STFT, raw waveform mixup, TFmask, volume perturbation weighted average SSAST traind on AudioSet data 4 simulation of anomalous samples, pre-trained model
78 Jinhyuk_SNU_task2_1 JinhyukSNU2022 AE 585648 log-mel energies
65 Hu_NJU_task2_1 HuNJU2022 VAE, CNN, ensemble, KNN 2489877 log-mel energies, spectrogram, raw waveform mixup weighted average 2
28 Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2022 CNN, GMM, ensemble 977811200 log-mel energies, magnitude spectrum mixup sum 40
30 Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2022 CNN, GMM, ensemble 977811200 log-mel energies, magnitude spectrum mixup sum 40
29 Wei_HEU_task2_1 WeiHEU2022 CNN, ArcFace 1347392 log-mel spectrogram, raw waveform
14 Wei_HEU_task2_2 WeiHEU2022 clustering, ensemble 0 log-mel spectrogram average 2
17 Wei_HEU_task2_3 WeiHEU2022 CNN, ArcFace, clustering, ensemble 1347392 log-mel spectrogram, raw waveform average 2
22 Guan_HEU_task2_1 GuanHEU2022 GMM 33024 log-mel spectrogram
19 Guan_HEU_task2_2 GuanHEU2022 GMM 33024 log-mel spectrogram SMOTE
11 Guan_HEU_task2_3 GuanHEU2022 GMM, CNN, ensemble 4204083 log-mel spectrogram, raw waveform weighted 4
6 Guan_HEU_task2_4 GuanHEU2022 GMM, CNN, ensemble 4204083 log-mel spectrogram, raw waveform SMOTE weighted 4
59 Li_CTRI_task2_1 LiCTRI2022 AE, K-NN 74779233 log-mel energies maximum ResNet38, MobileNetV2 3 pre-trained model
15 Morita_SECOM_task2_1 MoritaSECOM2022 CNN, k-NN 994048 spectrogram
25 Morita_SECOM_task2_2 MoritaSECOM2022 CNN, k-NN 994048 PCEN
21 Morita_SECOM_task2_3 MoritaSECOM2022 CNN, k-NN 994048 spectrogram, PCEN, HPSS
16 Morita_SECOM_task2_4 MoritaSECOM2022 CNN, k-NN, LOF, GMM 994048 spectrogram, PCEN, HPSS
53 Yamashita_GU_task2_1 YamashitaGU2022 IDNN, CNN 5602001 log spectrogram
61 Yamashita_GU_task2_2 YamashitaGU2022 U-Net 2164433 log spectrogram
39 Yamashita_GU_task2_3 YamashitaGU2022 AE,IDNN,CNN,U-Net,ensemble 11201923 log spectrogram 3
48 Yamashita_GU_task2_4 YamashitaGU2022 AE,IDNN,CNN,U-Net,ensemble 13930196 log spectrogram, log-mel energies 4
67 CHO_SG_task2_1 CHOSG2022 CNN, ArcFace 907520 log-mel energies, raw waveform mixup
81 CHO_SG_task2_2 CHOSG2022 CNN, ArcFace 907520 log-mel energies, raw waveform mixup
83 Li_JAIST_task2 LiJAIST2022 AE 269992 temporal modulation features on the gammatone auditory filterban
56 Gou_UESTC_task2_1 GouUESTC2022 CNN, LOF 28579432 spectrogram
58 Gou_UESTC_task2_2 GouUESTC2022 CNN, LOF 28579432 spectrogram
60 Gou_UESTC_task2_3 GouUESTC2022 CNN, LOF 28579432 spectrogram
51 Gou_UESTC_task2_4 GouUESTC2022 AE, CNN, LOF 28847360 spectrogram, log-mel energies
57 PENG_NJUPT_task2_1 PENGNJUPT2022 MobileNetV2 713910 Fast spectral coherence
62 PENG_NJUPT_task2_2 PENGNJUPT2022 MobileNetV2 713910 Fast spectral coherence,Wavelet packet energy,log-Mel
32 Nejjar_ETH_task2_1 NejjarETH2022 CNN, k-NN 1453440 log-mel energies mixup
54 Nejjar_ETH_task2_2 NejjarETH2022 CNN, k-NN 1453440 log-mel energies mixup
27 Verbitskiy_DS_task2_1 VerbitskiyDS2022 CNN, ArcFace, k-NN 6009730 log-mel energies temporal cropping, SpecAgment
40 Verbitskiy_DS_task2_2 VerbitskiyDS2022 CNN, ArcFace, k-NN 6009730 log-mel energies, MFCC, GFCC temporal cropping, SpecAgment
34 Verbitskiy_DS_task2_3 VerbitskiyDS2022 CNN, ArcFace, k-NN, ensemble 12019460 log-mel energies, MFCC temporal cropping, SpecAgment average 2
31 Verbitskiy_DS_task2_4 VerbitskiyDS2022 CNN, ArcFace, k-NN, ensemble 18029190 log-mel energies, MFCC temporal cropping, SpecAgment average 3
64 Cohen_Technion_task2_1 CohenTechnion2022 AE, Spectral Clustering, OCSVM 269992 log-mel enegries
63 Cohen_Technion_task2_2 CohenTechnion2022 Spectral Clustering, OCSVM 861 MFCC
9 Deng_THU_task2_1 DengTHU2022 ensemble 192720000 spectrogram, log-mel energies average 6 AudioSet
18 Deng_THU_task2_2 DengTHU2022 ensemble 192720000 spectrogram, log-mel energies average 6
13 Deng_THU_task2_3 DengTHU2022 ensemble 97130000 spectrogram, log-mel energies average 5
7 Deng_THU_task2_4 DengTHU2022 ensemble 1320000 spectrogram, log-mel energies average 2
55 Liu_BUPT_task2_1 LiuBUPT2022 CNN 1316042 log-mel energies mixup average
49 Liu_BUPT_task2_2 LiuBUPT2022 CNN, normalizing flow 1316042 log-mel energies mixup average OpenL3
80 Kazakova_ITMO_task2_1 KazakovaITMO2022 AE 101270 mel spectrogram
86 Kazakova_ITMO_task2_2 KazakovaITMO2022 MobileNetV2 590000 STFT TimeStretch, PitchShift pre-trained model
85 Kazakova_ITMO_task2_3 KazakovaITMO2022 MobileNetV2, AE 691270 STFT, mel spectrogram TimeStretch, PitchShift pre-trained model
74 Kodua_ITMO_task2_1 KoduaITMO2022 CNN 4796303 raw waveform PANNS MobileNetV1 pre-trained model
2 Liu_CQUPT_task2_1 LiuCQUPT2022 CNN 1215430 log-mel energies HPSS
4 Liu_CQUPT_task2_2 LiuCQUPT2022 CNN 1215430 log-mel energies HPSS
3 Liu_CQUPT_task2_3 LiuCQUPT2022 CNN 1215430 log-mel energies HPSS
1 Liu_CQUPT_task2_4 LiuCQUPT2022 CNN 1215430 log-mel energies HPSS
79 Siang_NTHU_task2_1 SiangNTHU2022 Time-dilated CNN, self-attention 653860 log-mel spectrogram frequency masking
38 Tozicka_NSW_task2_1 TozickaNSW2022 AE, energy, LOF 7946361 raw waveform, spectrogram OpenL3 2 pre-trained OpenL3
41 Tozicka_NSW_task2_2 TozickaNSW2022 AE, energy, LOF 8042361 raw waveform, spectrogram OpenL3 2 pre-trained OpenL3
46 Tozicka_NSW_task2_3 TozickaNSW2022 AE, energy, KNN 7946361 raw waveform, spectrogram OpenL3 2 pre-trained OpenL3
43 Tozicka_NSW_task2_4 TozickaNSW2022 AE, energy, LOF, KNN 24420841 raw waveform, spectrogram OpenL3 4 pre-trained OpenL3
50 Almudevar_UZ_task2_1 AlmudevarUZ2022 Transformer, ArcFace, k-NN 21797760 log-mel energies mixup
45 Almudevar_UZ_task2_2 AlmudevarUZ2022 Transformer, ArcFace, k-NN 21797760 log-mel energies mixup
47 Almudevar_UZ_task2_3 AlmudevarUZ2022 Transformer, ArcFace, k-NN 21797760 log-mel energies mixup
77 Jalalia_AIT_task2_1 JalaliaAIT2022 AE 188680 log-mel energies
84 Zorin_AIRI_task2_1 ZorinAIRI2022 LOF log-mel energies mixup, random crop, resize
23 Venkatesh_MERL_task2_1 VenkateshMERL2022 CNN, k-NN 1062938 spectrogram
24 Venkatesh_MERL_task2_2 VenkateshMERL2022 CNN, AE, k-NN 1609938 spectrogram weighted average 2
8 Venkatesh_MERL_task2_3 VenkateshMERL2022 CNN, k-NN 1062938 spectrogram
10 Venkatesh_MERL_task2_4 VenkateshMERL2022 CNN, k-NN 1062938 spectrogram



Technical reports

Vision Transformer based embeddings extractor for Unsupervised Anomalous Sound Detection under Domain Generalization

Antonio Almudevar, Alfonso Ortega, Luis Vicente, Antonio Miguel, Eduardo Lleida
University of Zaragoza, Zaragoza, Spain

Abstract

Anomalous sound detection (ASD) is the task of identifying if a sound is normal or anomalous with respect to a given reference. In most scenarios, we have a large amount of normal data to design our model, but little or no anomalous data. When this situation occurs, the problem can be approached in an unsupervised manner, i.e., only normal data is used for design. In this report we present a solution for the DCASE2022 task 2 (Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques), which aims to address the ASD problem under domain generalization. This means that the data to develop the system belongs to the source domain, while the test data can belong to this domain or to a different one (target domain). The presented solution proposes an embeddings extractor based on a Vision Transformer (ViT) and makes use of the k-Nearest-Neighbor (k-NN) algorithm to obtain the anomaly score.

System characteristics
Classifier ArcFace, Transformer, k-NN
System complexity 21797760 parameters
Acoustic features log-mel energies
Data augmentation mixup
PDF

JLESS SUBMISSION TO DCASE2022 TASK2: BATCH MIXING STRATEGY BASED METHOD WITH ANOMALY DETECTOR FOR ANOMALOUS SOUND DETECTION

Jisheng Bai, Yafei Jia, Siwei Huang, Mou Wang, Jianfeng Chen
Joint Laboratory of Environmental Sound Sensing, School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, China

Abstract

Anomaly detection has a wide range of applications such as finding fraud cases in industry or indicating network intrusion in network security. Anomalous sound detection (ASD) for machine condition monitoring can detect anomalies in advance and prevent causing damage. However, the operational conditions of machines often change, leading to the different acoustic characteristics between training and test data. Domain generalization techniques are required to adapt the model to different conditions. In this paper, we present a self-supervised method for ASD using batch mixing strategy with margin loss and anomaly detector. The proposed batch mixing strategy randomly mixes the data from source and target domains in a mini-batch to adapt the model between different domains. Moreover, we adopt a self-supervised method using machine IDs with additive angular margin loss to extract acoustic representations. Finally, we use the acoustic representations to train anomaly detectors to detect anomalous sound. Experimental results on the development dataset of DCASE2022 taks2 show that our method outperforms the baseline systems.

System characteristics
Classifier CNN, LOF, Transformer, ensemble
System complexity 567110, 995590 parameters
Acoustic features log-mel energies, spectrogram
Data augmentation fmix, mixup
Decision making vote
Subsystem count 3
PDF

Self-Supervised Learning Methods using ST-gram for Anomaly Machine Sound Detection

Wonki Cho
Dept of Computer Engineering, Sogang University, Seoul, Republic of Korea

Abstract

It is difficult to apply supervised learning to anomaly detection due to absence of abnormal data. Therefore, in anomaly detection, Therefore, we use unsupervised anomaly detection, which assumes that most of the data is a normal sample and learns without obtaining a label. In this paper, A self-supervised learning method is proposed for unsupervised anomaly detection. This network performs self-supervised classification using metadata associated with the audio files and compare to labeled normal and abnormal. To better extract the characteristics of the machine sounds, We adopt ST-gram for Spectral-temporal feature fusion and compared performance with some CNN Networks for DCASE 2022 Challenge task2: Unsupervised Anomaly Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques. As a result, Our method wasn’t performed well except ’Slider’ machine sounds.

System characteristics
Classifier ArcFace, CNN
System complexity 907520 parameters
Acoustic features log-mel energies, raw waveform
Data augmentation mixup
PDF

UNSUPERVISED ANOMALOUS DETECTION BASED ON RIEMANNIAN GEOMETRY

Or Cohen, Yahav Vinokur, Asaf Arad, Dolev Vaknin, Shahaf-Yaron Peleg, Alon Amar
The Andrew and Erna Viterbi Faculty of Electrical and Computer Engineering, Technion, Israel Institute of Technology., Haifa, Israel and Acoustics Research Center, Haifa, Israel

Abstract

This technical report presents our proposed algorithms for the task 2 of the DCASE2022 challenge, which is unsupervised anomalous sound detection for machine condition monitoring by applying domain generalization techniques. We suggest two methods for feature extraction. The first method is based on extracting features using the latent space of an Autoencoder, and the second method is based on using the Mel-frequency cepstral coefficients (MFCC) to represent the signal. We represent the features using symmetric positive-definite (SPD) matrices. As there maybe a domain shift between the train data and the test data, we first performed spectral clustering given the Riemannian distances between the SPD matrices. A one class SVM is then trained on each of the centers of the clusters and is used to detect the anomalies in the data.

System characteristics
Classifier AE, OCSVM, Spectral Clustering
System complexity 269992, 861 parameters
Acoustic features MFCC, log-mel enegries
PDF

Description and Discussion on DCASE 2022 Challenge Task 2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques

Kota Dohi, Keisuke Imoto, Noboru Harada, Daisuke Niizumi, Yuma Koizumi, Tomoya Nishida, Harsh Purohit, Takashi Endo, Masaaki Yamamoto, Yohei Kawaguchi
Research and Development Group, Hitachi, Ltd., Tokyo, Japan and Doshisha University, Kyoto, Japan and Google LLC, Tokyo, Japan and NTT Communication Science Labs, Kanagawa, Japan

Abstract

We present the task description of the Detection and Classification of Acoustic Scenes and Events (DCASE) 2022 Challenge Task 2: "Unsupervised anomalous sound detection (ASD) for machine condition monitoring applying domain generalization techniques". Domain shifts are a critical problem for the application of ASD systems. Because domain shifts can change the acoustic characteristics of data, a model trained in a source domain performs poorly for a target domain. In DCASE 2021 Challenge Task 2, we organized an ASD task for handling domain shifts. In this task, it was assumed that the occurrences of domain shifts are known. However, in practice, the domain of each sample may not be given, and the domain shifts can occur implicitly. In 2022 Task 2, we focus on domain generalization techniques that detects anomalies regardless of the domain shifts. Specifically, the domain of each sample is not given in the test data and only one threshold is allowed for all domains.

System characteristics
Classifier AE, MNV2
System complexity 269992 parameters
Acoustic features log-mel energies
PDF

AITHU SYSTEM FOR UNSUPERVISED ANOMALOUS DETECTION OF MACHINE WORKING STATUS VIA SOUNDING

Jia Liu, Yufeng Deng, Anbai Jiang, Yuchen Duan, Jitao Ma, Xuchu Chen, Pingyi Fan, Cheng Lu, Wei-Qiang Zhang
Department of Electronic Engineering, Tsinghua University, Beijing, China and Tsinghua University, Beijing, China

Abstract

This report describes the AITHU system for the DCASE 2022 Challenge Task 2, which aims to detect anomalous machine status via sounding by using machine learning methods, where the training dataset itself does not contain any examples of anomalies. We build six subsystems, including three self-supervised classification methods, two probabilistic methods and one generative adversarial network (GAN) based method. Our final submission are four ensemble systems, which are different combinations of the six subsystems. The best official score of the ensemble systems can achieve 86.81% on the development dataset, whereas the corresponding Autoencoder-based baseline and the MobileNetV2-based baseline are with scores of 52.61% and 56.01%, respectively.

System characteristics
Classifier ensemble
System complexity 1320000, 192720000, 97130000 parameters
Acoustic features log-mel energies, spectrogram
Decision making average
Subsystem count 2, 5, 6
External data usage AudioSet
PDF

ENSEMBLE OF MULTIPLE ANOMALY DETECTORS UNDER DOMAIN GENERALIZATION CONDITIONS

Shuxian Wang, Yajian Wang, Diyuan Liu, Fan Chu, Yunqing Li, Jia Pan, Jun Du, Tian Gao, Qing Wang
Department of Electronic Engineering and Information Science, University of Science and Technology of China, Hefei, China and iFLYTEK, Hefei, China and University of Science and Technology of China, Hefei, China

Abstract

This technical report outlines our solution to DCASE 2022 Challenge Task 2, Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques. The goal is to detect recordings that contain anomalous machine sounds in the test set using only normal sound data in the training set. Our approaches are based on an ensemble of a self-supervised classifier model, an autoencoder, a binary classification model that utilizes task irrelevant outliers as pseudo-anomalous data and a distance metric based model.

System characteristics
Classifier AE, Section ID classification, binary classification, cosine distance, ensemble
System complexity 182M, 183M, 184M, 185M parameters
Acoustic features STFT, log-mel energies, raw waveform
Data augmentation mixup, TFmask, volume perturbation
Decision making weighted average
System embeddings SSAST traind on AudioSet data
Subsystem count 4
External data usage simulation of anomalous samples, pre-trained model
PDF

UNSUPERVISED ANOMALOUS SOUND DETECTION USING FEATURE EXTRACTOR AND ANOMALY DETECTOR

Jiacheng Gou, Chuang Shi, Huiyong Li
School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China and University of Electronic Science and Technology of China, Chengdu, China

Abstract

This report proposes an anomalous sound detection method based on feature extraction and anomaly detection for DCASE 2022 task 2. In order to recognize the anomaly sound when only the normal sound is used as the training data, we use the clip of the spectrogram and the corresponding section name to train a feature extractor to generate the features of the normal sound. Then the anomaly detector is used to calculate the intensity of anomaly between the test sound features and the normal sound features, to provide the anomaly score of the test sound. In view of the domain generalization, the source domain and target domain select different shifts when clipping spectrum, and select different anomaly detectors based on whether the sound belongs to the source domain or target domain.

System characteristics
Classifier AE, CNN, LOF
System complexity 28579432, 28847360 parameters
Acoustic features log-mel energies, spectrogram
PDF

The DCASE2022 Challenge Task 2 System: Anomalous Sound Detection with Self-supervised Attribute Classification and GMM-based Clustering

Feiyang Xiao, Youde Liu, Jian Guan, Yuming Wei, Qiaoxi Zhu, Tieran Zheng, Jiqing Han
College of Computer Science and Technology, Harbin Engineering University, Harbin, China and Harbin Institute of Technology, Harbin, China and Harbin Engineering University, Harbin, China and University of Technology Sydney, Ultimo, Australia

Abstract

This report describes our submission for DCASE2022 Challenge Task 2, an ensemble system for unsupervised anomalous sound detection (ASD), allowing domain shifts. It integrates two domain generalization methods, a self-supervised attribute classification and a GMM-based clustering for unsupervised ASD. Experiments were conducted on the development dataset of DCASE2022 Challenge Task 2. The results show that our ensemble system can achieve 88.5% in average AUC under the source domain, 78.5% in average AUC under the target domain, and 68.8% in average pAUC.

System characteristics
Classifier CNN, GMM, ensemble
System complexity 33024, 4204083 parameters
Acoustic features log-mel spectrogram, raw waveform
Data augmentation SMOTE
Decision making weighted
Subsystem count 4
PDF

AN ENSEMBLE METHOD FOR UNSUPERVISED ANOMALOUS SOUND DETECTION

Qinwen Hu, Kai Chen, Jing Lu
Key Laboratory of Modern Acoustics,, Nanjing University, Nanjing, China and Nanjing University, Nanjing, China

Abstract

This report describes our submitted system for DCASE2022 challenge task2 (Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques) [1] in detail. The system is composed of two modules, a hierarchical recurrent variational autoencoder and a self-supervised classifier, and the final score is a weighted average over the normalized results of the two systems. The anomaly scores are all calculated in the latent/embedding space.

System characteristics
Classifier CNN, KNN, VAE, ensemble
System complexity 2489877 parameters
Acoustic features log-mel energies, raw waveform, spectrogram
Data augmentation mixup
Decision making weighted average
Subsystem count 2
PDF

DCASE CHALLENGE 2022, TASK 2: VARIATIONAL DENSE AUTOENCODER FOR UNSUPERVISED ANOMALOUS SOUND DETECTION OF MACHINERY

Anahid Jalali, Lam Pham, Clemens Heistracher, Denis Katic, Alexander Schindler
Data Science and Artificial Intelligence, Austrian Institute of Technology (AIT), Vienna, Austria

Abstract

In this study, we present an unsupervised anomalous sound detection framework trained on the DCASE2022 audio dataset. We use variational dense autoencoder to reconstruct the machine’s healthy (normal) state and use the reconstruction loss as a threshold for detecting the anomalies in an unsupervised manner. Our framework outperforms DCASE2021 benchmarks in target domains. The dense autoencoder has a harmonic mean of AUC of 72.10% (source), and 45.49% (target) and pAUC of 54.09%. Our framework achieved the harmonic mean AUC of 68.78 and pAUC of 53.96, over all the machines. Our target domain arithmetic average, however, achieved 47.77% (baseline: 45.49%) which shows an slight improved performance from the dense autoencoder.

System characteristics
Classifier AE
System complexity 188680 parameters
Acoustic features log-mel energies
PDF

ENSEMBLE METHOD FOR UNSUPERVISED ANOMALOUS SOUND DETECTION

Jinhyuk Park, Sangwoong Yoon, Yonghyeon Lee, Minjun Son, Frank C. Park
Robotics Laboratory, Seoul National University, Seoul, South Korea

Abstract

We propose an anomalous sound detection method for DCASE2022 Challenge Task2. The method is basically an ensemble of multiple autoencoder-based approaches. The model reconstruct the input Mel spectrogram and decide it is an anomaly if the reconstruction error is higher than a threshold. The area under curve (AUC) performance achieved by the proposed approach is 53.35% on source domain and 43.48% on target domain, and the partial AUC is 48.00%.

System characteristics
Classifier AE
System complexity 585648 parameters
Acoustic features log-mel energies
PDF

ANOMALOUS SOUND DETECTION WITH AUTOENCODER AND IMAGE MOBILENET USING OUTLIER EXPOSURE APPROACH

Sophia Kazakova, Andrey Semenov, Andrey Surkov, Sergei Astapov
ITMO University, St. Petersburg, Russia

Abstract

This technical report describes the autoencoder and MobileNetV2-based approach for DCASE 2022 Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques task [1]. Firstly, a basic Autoencoder-based architecture was developed. Then the outlier exposure approach was tested on DCASE 2022 Challenge Task 2 Development Dataset [2]. Proven its effectiveness, it then was used as a part of an image MobileNetv2 system. To tackle the challenge of domain shift and make the dataset more balanced in terms of source/target classes we used data augmentation with TimeStretch and PitchShift. Audio files then were transformed with STFT and saved as images. The MobileNetv2-based architecture was fine-tuned with those spectrograms and used for anomaly detection.

System characteristics
Classifier AE, MobileNetV2
System complexity 101270, 590000, 691270 parameters
Acoustic features STFT, mel spectrogram
Data augmentation TimeStretch, PitchShift
External data usage pre-trained model
PDF

ANOMALOUS SOUND DETECTION WITH PANNS MOBILENETV1 EMBEDDINGS

Ilya Kodua, Sophia Kazakova, Andrey Semenov
ITMO University, St. Petersburg, Russia

Abstract

This technical report describes the PANNs MobileNetv1-based approach for DCASE 2022 Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques task [1]. The objective of this task is to determine whether the sound emitted from the target machine class is normal or anomalous while having only the normal data for training purposes. We extract embeddings using external PANNs MobileNetV1 pre-trained model [2]. For anomaly score assignment we concatenate the embeddings obtained and then calculate the cosine distance to the first nearest neighbor in the embedding space for this sample’s class and section. For this report the GitHub code is available. PANNs embedding extraction is [3], and anomaly score calculation is [4].

System characteristics
Classifier CNN
System complexity 4796303 parameters
Acoustic features raw waveform
System embeddings PANNS MobileNetV1
External data usage pre-trained model
PDF

Two-stage anomalous sound detection systems using domain generalization and specialization techniques

Ibuki Kuroyanagi, Tomoki Hayashi, Kazuya Takeda, Tomoki Toda
Nagoya University and Human Dataware Lab. Co., Ltd., Nagoya, Japan and Human Dataware Lab. Co., Ltd. and Nagoya University, Nagoya, Japan and Nagoya University, Nagoya, Japan

Abstract

This report proposes anomalous sound detection (ASD) methods using domain generalization and specialization techniques for the DCASE 2022 Challenge Task 2. We propose two-stage ASD systems consisting of an outlier exposure-based feature extractor and an inlier modeling-based anomalous detector in serial. We further employ two approaches to deal with domain shift: a domain generalization approach and a domain specialization approach. Each approach improves performance significantly by adding several techniques to the two-stage ASD systems, such as generating pseudo-target domain data by Mixup and utilizing pseudo-anomalous data from Audioset. Our final systems are obtained by ensembling several systems with several hyperparameters for each approach. The proposed systems achieve 81.15 % in the harmonic mean of all machine types, sections, and domains for the area under the curve (AUC) and partial AUC (p = 0.1) on the development set.

System characteristics
Classifier CNN, Conformer, GMM, KNN, LOF, Transformer
System complexity 48789168 parameters
Acoustic features spectrogram
Data augmentation mixup,gaussian noise
Decision making average
System embeddings PyTorch Image Models
Subsystem count 20
External data usage pre-trained model,Audioseet
PDF

ANOMALOUS SOUND DETECTION USING CONTRASTIVE LEARNING

Seunghyeon Shin, Seokjin Lee
School of Electronic and Electrical Engineering, Kyungpook National University, Daegu, Republic of Korea

Abstract

We propose an unsupervised anomalous sound detection system for DCASE 2022 Task 2. We use self supervised contrastive learning with data augmentation as a feature extractor network. We use three kinds of data augmentation methods for contrastive learning. Then k-Nearest Neighbors are used to compute anomalous scores from extracted feature vectors. As a result, we show the detection performance of 88.58% in Area under Curve(AUC) and 74.40% in partial AUC(pAUC) with hyperparameter fixed.

System characteristics
Classifier Contrastive learning, k-NN
System complexity 11496000 parameters
Acoustic features log-mel energies
Data augmentation Harmonics modification, Temporal masking, F0 masking,
PDF

ANOMALOUS SOUND DETECTION WITH ENSEMBLE OF CNN-BASED FEATURES AND AUTOENCODER APPROACHES

Xiaoyu Li, Jie Yang, Hao Shen
Department of Big Data and Artificial Intelligence, China Telecom Corporation Research Institute, Beijing, China

Abstract

This paper introduces a solution with the ensemble of three anomalous sound detection (ASD) methods for the DCASE2022 Challenge Task 2[1] [2 [3]. This task is required to detect unknown anomalous sound basing on normal sound data. The first ASD method is using the audio clip of the machine which is normal, and the section index of audio clip to train the Convolutional Neural Network (CNN). Then, anomalous sound is detected by using feature vectors extracted from CNN. The second ASD method is an OE-based detector that uses MobileNetV2. The third ASD method is an IM-based detector that uses autoencoder (AE). As a result, our method achieves a harmonic mean of 72.70% over of area under the curve (AUC), and 60.35% in partial AUC (pAUC).

System characteristics
Classifier AE, K-NN
System complexity 74779233 parameters
Acoustic features log-mel energies
Decision making maximum
System embeddings ResNet38, MobileNetV2
Subsystem count 3
External data usage pre-trained model
PDF

Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Using Temporal Modulation Features on Gammatone Auditory Filterbank

Kai Li, Quoc-Huy Nguyen, Yasuji Ota, Masashi Unoki
School of Information Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa, 923-1292, Japan and Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa, 923-1292, Japan

Abstract

Anomalous sound detection (ASD) is a task to identify whether the sound emitted from a target machine is normal or not. Subjectively, timbral attributes, such as sharpness and roughness, are crucial for human beings to distinguish anomalous and normal sounds. However, the feature frequently used in existing methods for ASD is the log-mel-spectrogram, which cannot capture information in the time domain. This paper proposes an ASD method using temporal modulation features on the gammatone auditory filterbank (TMGF) to provide temporal characteristics for machine-learning-based methods. We evaluated the proposed method using the area under the ROC curve (AUC) and the partial area under the ROC curve (pAUC) with sounds recorded from seven kinds of machines. Compared with the baseline method of the DCASE2022 challenge, the proposed method provides a better ability for domain generalization, especially for machine sounds recorded from the valve.

System characteristics
Classifier AE
System complexity 269992 parameters
Acoustic features temporal modulation features on the gammatone auditory filterban
PDF

Unsupervised Anomalous Sound Detection Under Domain Shift Conditions Based on MobileFaceNets and Masked Autoregressive Flow

Gang Liu, Yi Liu, Shifang Cai, Minghang Chen
School of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing, China and Beijing University of Posts and Telecommunications, Beijing, China

Abstract

We present our submission to the DCASE2022 Challenge Task 2, which aims to promote research in unsupervised anomalous sound detection under domain shift condition. We propose two architectures to solve this problem, one is a self-supervised model adopting MobileFaceNets, and the other is one density estimation probability distribution model based on Masked Autoregressive Flow.

System characteristics
Classifier CNN, normalizing flow
System complexity 1316042 parameters
Acoustic features log-mel energies
Data augmentation mixup
Decision making average
System embeddings OpenL3
PDF

ROBUST ANOMALY SOUND DETECTION FRAMEWORK FOR MACHINE CONDITION MONITORING

Ying Zeng, Hongqing Liu, Lihua Xue, Yi Zhou, Lu Gan
Chongqing University of Posts and Telecommunications and AI Lab, Xiaomi Corporation, Chongqing and Beijing, China and Chongqing University of Posts and Telecommunications, Chongqing, China and Brunel University, London UB8 3PH, U.K.

Abstract

This technical report describes our team’s submission to DCASE 2022 Task 2. In this report, we propose a robust training framework for anomalous sound detection, which includes feature preprocessing, model pretraining, joint loss, and anomaly score selection. The experimental results show that our anomalous sound detection model outperforms the official model, with an average performance improvement of 22.08% based on the official scoring method.

System characteristics
Classifier CNN
System complexity 1215430 parameters
Acoustic features log-mel energies
Front end system HPSS
PDF

COMPARATIVE EXPERIMENTS ON SPECTROGRAM REPRESENTATION FOR ANOMALOUS SOUND DETECTION

Kazuki Morita, Tomohiko Yano, Khai Q. Tran
Intelligent Systems Laboratory, SECOM CO.,LTD., Tokyo, Japan

Abstract

In this paper, we propose an anomalous sound detection method for DCASE2022task2. This is the task of anomalous sound detection for machine condition monitoring, and it is required to detect unknown anomalous sound only from normal sound data. Our system is based on a submission system for DCASE2021task2, and we newly evaluated variations in the time-frequency representation used in anomalous sound detection. As a result, the proposed method showed a detection performance of 84.80% for source domain and 82.26% for target domain in Area Under Curve (AUC) and 68.65% in partial AUC (pAUC).

System characteristics
Classifier CNN, GMM, LOF, k-NN
System complexity 994048 parameters
Acoustic features HPSS, PCEN, spectrogram
PDF

CNN-BASED ANOMALOUS SOUND DETECTION SYSTEM FOR DOMAIN GENERALIZATION

Hiroki Narita, Akira Tamamori
Aichi Institute of Technology, Aichi, Japan

Abstract

This paper is a technical report for DCASE Challenge 2022 Task 2. Our submitted model consists of a self-supervised CNN model that predicts attribute information. We have ensembled three models but have not changed the architecture, and have achieved performance improvement only by changing the data expansion, training method, and anomaly detection method. Self-supervised learning with label information has been a powerful method in previous anomaly detection competitions, and we argue that it is equally powerful in this competition.

System characteristics
Classifier EfficientNet-B1, Mahalanobis
System complexity 23382552, 7794184 parameters
Acoustic features log-mel energies
Data augmentation Mixup, Frequency Masking, Mixup, Time Masking, Frequency Masking, Mixup, Time Masking, Frequency Masking, SevenBandParametricEQ
System embeddings PyTorch Image Models (EfficientNet-B1)
Subsystem count 3
External data usage pre-trained model
PDF

DCASE CHALLENGE 2022: SELF-SUPERVISED LEARNING PRE-TRAINING, TRAINING FOR UNSUPERVISED ANOMALOUS SOUND DETECTION

Ismail Nejjar, Jean Meunier-Pion, Gaetan Frusque, Olga Fink
IMOS, ETHZ & EPFL, Swizerland and EPFL, Lausanne, Swizerland

Abstract

This technical report presents our proposed approaches for Task 2 of the DCASE 2022 Challenge, Unsupervised anomalous sound detection (ASD) for machine condition monitoring by applying domain generalization techniques. The main objective of this challenge is to detect anomalous machine sounds regardless of the domain shifts. Our approach introduces a two-step learning process, where normal sounds of each specific machine type are used to pretrain a Convolutional Neural Network (CNN) in a self-supervised way. Three objectives are thereby pursued: (1) reveal the impact of attributes on the data by enforcing embeddings in the same batch to be different (2) obtain uncorrelated embedding features containing specific information, (3) respecting defined geometrical constraints between the different domains. The model trained in an unsupervised way is then fine-tuned on the labels of the section indices. Ultimately, anomalous sounds are detected by using the feature vectors extracted from the CNN and applying k-NN to them. As a result, for the development set, it is shown that the presented framework significantly outperforms both baselines.

System characteristics
Classifier CNN, k-NN
System complexity 1453440 parameters
Acoustic features log-mel energies
Data augmentation mixup
PDF

UNSUPERVISED ABNORMAL SOUND DETECTION BASED ON SPECTRAL COHERENCE AND FEATURE FUSION IN DOMAIN DISPLACEMENT CONDITION

Tao Peng, Rui Qiu, Junyi Zhu, Yao Xiao, Su Wang, Yipeng Zhang, Chenyang Zhu, Shengchen Li, Xi Shao
Telecommunications & Information Engineering, Nanjing, China and School of Advanced Technology, Suzhou, China

Abstract

The DCASE2022 Challenge Task2 is to develop an unsupervised detection system of anomalous sounds for seven types of machines under domain shifted conditions. In this paper, two systems are proposed: one only uses spectral coherence as feature input and another combines spectral coherence, wavelet and log Mel. It shows that three-feature fusion has significantly improved the results compared with the baseline in general, but sometimes spectral coherence alone can lead to better results. Therefore, we suggest to use both methods in order to get stable results.

System characteristics
Classifier MobileNetV2
System complexity 713910 parameters
Acoustic features Fast spectral coherence, Wavelet packet energy, log-Mel
PDF

OUTLIER-AUGMENTED CONTRASTIVE CLUSTERING FOR ANOMALY SOUND DETECTION WITH UNBALANCED DOMAIN

You-Siang Chen, Mingsian R. Bai
Power Mechanical Engineering, National Tsing Hua University, Hsinchu, Taiwan and National Tsing Hua University, Hsinchu, Taiwan

Abstract

In this report, we developed a deep neural network (DNN) that can perform the deep clustering for the embedding vectors of machine sounds. The time-dilated convolutional neural network (TDCN) with attention mechanism was exploited to extract important features related to the time sequence. In addition, frequency masking is applied to the non-target sections of the machine sound to further increase the data size of the outliers. The results show that by applying the data augmentation to the outliers, the AUC performance can be improved. Furthermore, the deep clustering is able to contrastively attract and separate the machine sounds with unbalanced domain.

System characteristics
Classifier Time-dilated CNN, self-attention
System complexity 653860 parameters
Acoustic features log-mel spectrogram
Data augmentation frequency masking
PDF

DADAED - Double Anomaly Detector with AEDiff

Jan Tozicka, Marek Bezusek, Karel Durkota, Michal Linda
NeuronSW SE, Prague, Czech Republic

Abstract

This report describes our submissions to the DCASE 2022 challenge Task 2 "Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring under Domain Shifted Conditions." Acoustic-based machine condition monitoring is a challenging task with a very unbalanced training dataset. Moreover, due to domain-shift, testing data may come from a different distribution than the training data, which makes the task even more difficult. In this submission, we propose two novel extensions of anomaly detection based on the reconstruction of auto-encoder (AE) network. The first approach uses the raw difference between AE input and its reconstructed output (instead of typical reconstruction error based anomaly detectors). The second approach extends the first approach with an additional anomaly score of autoencoder’s latent vectors. The combination of these two anomaly scores is then used to determine the final anomaly score.

System characteristics
Classifier AE, KNN, LOF, energy
System complexity 24420841, 7946361, 8042361 parameters
Acoustic features raw waveform, spectrogram
System embeddings OpenL3
Subsystem count 2, 4
External data usage pre-trained OpenL3
PDF

Disentangled surrogate task learning for improved domain generalization in unsupervised anomalous sound detection

Satvik Venkatesh, Gordon Wichern, Aswin Subramanian, Jonathan Le Roux
Speech & Audio Team, Mitsubishi Electric Research Laboratories (MERL), Cambridge, MA, USA and Mitsubishi Electric Research Laboratories (MERL), Cambridge, MA, USA

Abstract

We present our submission to the DCASE2022 Challenge Task 2, which focuses on domain generalization for anomalous sound detection. We investigated a novel multi-task learning framework that disentangles domain-shared features and domain-specific features. Disentanglement leads to better latent features and also increases flexibility in post-processing due to the availability of multiple embedding spaces. Our disentangled model obtains an overall harmonic mean of 74.57% on the development set, surpassing the MobileNetV2 baseline, which obtains 56.01%. Lastly, we explore the use of machine-specific loss functions and domain generalization methods, which improves our overall performance to 76.42%.

System characteristics
Classifier AE, CNN, k-NN
System complexity 1062938, 1609938 parameters
Acoustic features spectrogram
Decision making weighted average
Subsystem count 2
PDF

Unsupervised Anomalous Sound Detection Using Multiple Time-Frequency Representations

Sergey Verbitskiy, Milana Shkhanukova, Viacheslav Vyshegorodtsev
Deepsound, Novosibirsk, Russia

Abstract

This technical report describes our approach for the DCASE2022 Challenge Task 2. This task aims to continue research on unsupervised anomalous sound detection and develop new high-performing systems for monitoring the condition of machines. In contrast to the DCASE2021 Challenge Task 2, the 2022 task primarily focuses on domain generalization. First and foremost, we propose the idea of using ensembles of 2D CNN-based systems that utilize different time-frequency representations as input features. We use normal sound clips and their section indices to train our anomalous sound detection (ASD) systems for each machine type, and embedding vectors extracted from our CNNs, cosine similarity, and the k-nearest neighbors algorithm (k-NN) to calculate the anomaly scores of test clips. As a result, our method achieves the official score of 0.725 on the development dataset and significantly outperforms the baseline systems.

System characteristics
Classifier ArcFace, CNN, ensemble, k-NN
System complexity 12019460, 18029190, 6009730 parameters
Acoustic features GFCC, MFCC, log-mel energies
Data augmentation temporal cropping, SpecAgment
Decision making average
Subsystem count 2, 3
PDF

Anomalous Sound Detection System with Self-challenge and Metric Evaluation for DCASE2022 Challenge Task 2

Yuming Wei, Jian Guan, Haiyan Lan, Wenwu Wang
College of Computer Science and Technology, Harbin Engineering University, Harbin, China and Centre for Vision, Speech and Signal Processing, University of Surrey, Guildford, UK

Abstract

This technical report describes our submission for DCASE2022 Challenge Task 2. To solve the domain generalization problem in anomalous sound detection (ASD), we present an ensemble system with two proposed unsupervised anomalous sound detection methods, i.e., a self-supervised classifier with a self-challenge strategy, and a distance metric evaluation based method. Experiments conducted show that our ensemble system can achieve an average of 87.07\% in harmonic mean AUC score under the source domain (h-mean AUC-s), and an average of 76.22\% in harmonic mean AUC score under the target domain (h-mean AUC-t), and an average of 66.76\% in harmonic mean pAUC (h-mean pAUC) score.

System characteristics
Classifier ArcFace, CNN, clustering, ensemble
System complexity 0, 1347392 parameters
Acoustic features log-mel spectrogram, raw waveform
Decision making average
Subsystem count 2
PDF

An Outlier Exposed Anomalous Sound Detection System for Domain Generalization in Machine Condition Monitoring

Kevin Wilkinghoff
Communication Systems, Fraunhofer Institute for Communication, Information Processing and Ergonomics FKIE, Wachtberg, Germany

Abstract

Emitted machine sounds can change drastically due to a change in settings of machines or due to varying noise conditions. This is a problem when monitoring the condition of these machines with a trained anomalous sound detection system because after changing the acoustic conditions the normal sounds are often falsely marked as anomalous. The goal of task 2 "Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques" of the DCASE 2022 challenge is to develop systems that reliably detect anomalous sounds regardless of whether characteristics of machine sounds are changing or not. In this work, a conceptually simple outlier exposed anomalous sound detection system is presented that is specifically designed for domain generalization. To this end, multiple feature representations and carefully designed sub-system architectures are utilized inside a single neural network. Furthermore, a technique called domain mixup is presented to further improve the domain generalization capabilities.

System characteristics
Classifier CNN, GMM, ensemble
System complexity 977811200 parameters
Acoustic features log-mel energies, magnitude spectrum
Data augmentation mixup
Decision making sum
Subsystem count 40
PDF

ANOMALY DETECTION USING AUTOENCODER, IDNN AND U-NET USING ENSEMBLE

Jun’ya Yamashita, Ryosuke Tanaka, Keisuke Ikeda, Shiiya Aoyama, Satoru Hayamizu, Satoshi Tamura
Gifu University, Gifu, Japan

Abstract

This paper presents our efforts for DCASE 2022 Challenge Task 2. We built several anomaly detectors based on AutoEncoder (AE), Interpolation Deep Neural Network (IDNN) with acoustic noise, U-Net with mask patches. Through experiments using those detection schemes as well as training and development data sets, we found the best model for each machine type is different. We further integrated anomaly scores obtained from every detectors by ensemble technique. Our results show that we could improve Area Under the Curve (AUC) scores particularly for target domains.

System characteristics
Classifier AE, CNN, IDNN, U-Net, ensemble
System complexity 11201923, 13930196, 2164433, 5602001 parameters
Acoustic features log spectrogram, log-mel energies
Subsystem count 3, 4
PDF

ANOMALY DETECTION WITH SELF-SUPERVISED AUDIO EMBEDDINGS

Ivan Zorin, Ilya Makarov
Industrial AI, Artificial Intelligence Research Institute, Moscow, Russia

Abstract

The majority of approaches to machine condition monitoring via anomalous sound detection are based on supervised learning. The metadata of the datasets is used as data labels for training supervised models. However, data labeling is expensive and often impossible for industries with significant amount of equipment. In this case self-supervised methods could solve the problem since they do not require labeled data. In this work we applied the recent self-supervised approach to compute embeddings of audio signals named BYOL-A and classical machine learning method Local Outlier Factor (LOF) to compute outlier scores for anomalous sounds. The main focus of this work is to not use any labels from the metadata of the datasets and explore a self-supervised learning approach.

System characteristics
Classifier LOF
Acoustic features log-mel energies
Data augmentation mixup, random crop, resize
PDF