Add callbacks to monitor training progress and prevent overfitting

tallamjr / astronet

Efficient Deep Learning for Real-time Classification of Astronomical Transients and Multivariate Time-series

Apache License 2.0

14 stars 3 forks source link

Add callbacks to monitor training progress and prevent overfitting #47

Closed tallamjr closed 3 years ago

tallamjr commented 3 years ago

Currently, hyperparameter optimisation and training take place for a given number of epochs, it would be desirable to be able to implement an early stopping criteria if it is seem to be overfitting, or that the progress has plateaued.

tensorflow.keras.callbacks seem to be a good fit for this.

See Coursera videos for details on how to implement

tallamjr commented 3 years ago

Closed with #50