/MTAD_GAT

PyTorch implementation of MTAD-GAT (Multivariate Time-Series Anomaly Detection via Graph Attention Network)

Primary LanguagePython

Loss

  • Train Loss
  • Valid Loss

Result

  • epsilon_threshold
precision recall f1
0.543145 1.000000 0.703946
  • pot_threshold
precision recall f1
0.641581 1.000000 0.781663
  • bestf1_threshold
precision recall f1
0.994087 0.998515 0.996296