search

lyn1874 / memAE

unofficial implementation of paper Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder (MemAE) for Unsupervised Anomaly Detection
69 stars 24 forks source link

train memory sparse loss #15

Open xinguo2 opened 2 years ago

xinguo2 commented 2 years ago

When I train the model with my custom data, the train memory sparse loss increases first and then decreases. could you tell me the reason?