MemAE for anomaly detection. -- Gong, Dong, et al. "Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detection". ICCV 2019.
Hi! I've been trying to replicate the MNIST experiment in the paper. I'm interest in this threshold too. In my experiments the reconstruction error doesn't see out of distribution for anomaly samples.
Thanks the great work! Could you tell me the threshold of the recon_error, which used to distinguish between normal and abnormal.