Open Josef23 opened 3 years ago
@Josef23 They're still unsupported but can you share more details about what you'd like Auto Encoders for?
@JakeRadMSFT Thank you for replaying. I am using autoencoders for denoising and potentially anomaly detection. Currently, I am using interprocess communications to get the results of a training/inference implementation in Python/Tensorflow to a c# environment. So you see why I would like to use ML.Net to do at least the inference only in c# (with the models converted to ONNX format). Also, I am starting to play around with GANs (obvious next step?) but, from looking around a bit, I think that these are also not supported?
After now some years, and the initially mentioned closed issue being mine too... I still think autoencoders would be good to have in ML.Net.
Anomaly Detection and Dimensionality reduction are my interests here.
I cannot find anything more relevant about autoencoders in ML.Net than this closed issue: https://github.com/dotnet/machinelearning.
Is it possible now to use autoencoders in ML.Net?