@tony-hong @sama25100 @SimonP07
I Implemented a LSTM in Tensorflow and trained it using season data of 50 doped and 50 clean riders with a 75/25 train test split. This is the result of 150 training epochs. I think it looks very promising!
In Tensorflow we can easily bias the loss function to account for the data imbalance. I already used this functionality here.
I will try to deploy the model on the computing cluster later in the evening
@tony-hong @sama25100 @SimonP07 I Implemented a LSTM in Tensorflow and trained it using season data of 50 doped and 50 clean riders with a 75/25 train test split. This is the result of 150 training epochs. I think it looks very promising!
In Tensorflow we can easily bias the loss function to account for the data imbalance. I already used this functionality here.
I will try to deploy the model on the computing cluster later in the evening