How to Choose the Best Source Model for Transfer Learning

  1. Truncate all of the source networks at the desired layer.
  2. Input the target data into each of the networks to get the “encodings”.
  3. Calculate how well the encodings cluster the target data using the Mean Silhouette Coefficient. The Silhouette Coefficient is a number between 1 and -1, where 1 indicates an optimal clustering and -1 indicates a poor clustering. For a more elaborate discussion of the Silhouette Coefficient see our paper (link at the end of the post). Other resources include Wikipedia and the original paper Silhouettes: A graphical aid to the interpretation and validation of cluster analysis by Peter Rousseeuw.
  4. Select the source model that yields the highest Silhouette Coefficient as the source model for Transfer Learning.
Our source model selection method

--

--

Love podcasts or audiobooks? Learn on the go with our new app.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store