異常検出
異常検出は、データセットの大部分から大幅に逸脱する不審なパターンや予期せぬパターンを特定することを目的とした二値分類タスクです。このタスクの目標は、これらの外れ値(エラーや詐欺、その他の異常事象を示す可能性がある)を見つけ出し、さらなる調査のためにマークすることです。異常検出は、金融リスク管理、サイバーセキュリティ、医療診断などの分野で重要な応用価値を持っています。
MVTec AD
GLASS
VisA
ReContrast
MVTec LOCO AD
CSAD
One-class CIFAR-10
CSI
CUHK Avenue
HF2VAD+SSPCAB
ShanghaiTech
SSMTL+UBnormal
UCR Anomaly Archive
Auto-Encoder with Regression (AER)
Fishyscapes L&F
cDNP+OE
MPDD
GLASS
One-class CIFAR-100
GeneralAD
BTAD
MuSc (zero-shot)
UBnormal
TimeSformer
UCSD Ped2
Background-Agnostic
Unlabeled CIFAR-10 vs CIFAR-100
CSI
Fashion-MNIST
One-class ImageNet-30
CSI
Numenta Anomaly Benchmark
HTM AL
Road Anomaly
RbA
AeBAD-S
MSFR
Anomaly Detection on Unlabeled CIFAR-10 vs LSUN (Fix)
Fishyscapes
RPL+CoroCL
AeBAD-V
MMR
Hyper-Kvasir Dataset
Leave-One-Class-Out CIFAR-10
Leave-One-Class-Out ImageNet-30
BCE-CLIP (OE)
MNIST
Anomaly Detection on Unlabeled ImageNet-30 vs CUB-200
Anomaly Detection on Anomaly Detection on Unlabeled ImageNet-30 vs Flowers-102
InsPLAD
AttentDifferNet (SENet-AlexNet)
LAG
CCD
Cats-and-Dogs
Self-Supervised One-class SVM, RBF kernel
DIOR
Self-Supervised One-class SVM, RBF kernel
Lost and Found
MVTEC AD textures
PHEVA
MPED-RNN
Surface Defect Saliency of Magnetic Tile
HETMM
Corridor
Two-stream
MVTec AD Textures Domain Generalization
FABLE
ODDS
kNN
UCSD Peds2
UEA time-series datasets
SINBAD
voraus-AD
MVT-Flow
MVTEC 3D-AD
CDO
PAD Dataset
SplatPose
Thyroid
RCALAD
Vehicle Claims
Random Forest
ADNI
Brainomaly
AG News
DATE
ASSIRA Cat Vs Dog
Shell-based Anomaly (supervisered)
BottleCap
DFC
Census
DevNet
CIFAR-10
COCO-OOC
Forest CoverType
IITB Corridor
Kaggle-Credit Card Fraud Dataset
kdd 99
PCA via oversampling
KDD Cup 1999
KSDD2
SAA+
MIT-BIH Arrhythmia Database
MNIST-test
OGNET
Musk v1
MVTec 3D-AD (RGB)
NB15-Analysis
NB15-Backdoor
NB15-DoS
DIF
Real 3D-AD
Reg 3D-AD
ShanghaiTech Campus
TSGAD
SMD
MAAT
STL-10
Shell-based Anomaly (supervised)
Street Scene
PGM
SVHN
RCALAD
TII-SSRC-23
UCF-Crime
WFDD
GLASS
MVTec-AD