Deep Bottleneck
doc/ruediger

Contents

  • Big Picture
  • Contributing
  • Glossary
  • Literature
  • Literature Summary
  • User guide
  • Experiments
    • Description of cohorts
    • Comparing activation functions for a minimal model
    • Calculation of mutual information for different parts of the dataset
    • Experiment Evaluation of Activation Functions
    • Standard vs. Weighted Binning
    • Effect of weight renormalization on activity patterns
    • The data set provided by Tishby
    • Our attempt to generate the data set above
  • Indices and tables
  • API Documentation
Deep Bottleneck
  • Docs »
  • Experiments
  • Edit on GitHub

Experiments¶

  • Description of cohorts
  • Comparing activation functions for a minimal model
  • Calculation of mutual information for different parts of the dataset
  • Experiment Evaluation of Activation Functions
    • 1. The Hypothesis
    • 2. Experimental Setting
    • 3. Results
  • Standard vs. Weighted Binning
    • In this notebook we try to show that the normal binning approach is not useful and try to derive a new one.
  • Effect of weight renormalization on activity patterns
    • Experiments with max_weight_norm=0.8
    • Experiment with max_weight_norm=0.4
    • Supplementary material
  • The data set provided by Tishby
  • Our attempt to generate the data set above

Indices and tables¶

  • Index
  • Module Index
  • Search Page
Next Previous

© Copyright 2018, Deep Bottleneck study project. Revision 208ab473.

Built with Sphinx using a theme provided by Read the Docs.