Closed jlro001 closed 5 years ago
Some Newer deep learning methods use reconstruction error using autoencoders instead of prediction error (used in this repo) You might want to read this paper for more information: Malhotra, Pankaj, et al. "LSTM-based encoder-decoder for multi-sensor anomaly detection." arXiv preprint arXiv:1607.00148 (2016).
I am very appreciated that the authors have contributed the codes. However, when I test the codes, I can only get about f-Beta=0.3 on these datasets. and I find out that is because the precision is very low, not bigger than 0.5. And I want to know if there are some tricks in the experiments?