Presentations for IEEE WCCI 2014

In the Presentations section, I have added the slides of two of the papers I presented:

  1. An Effective Criterion for Pruning Reservoir’s Connections in Echo State Networks – We present a new criterion for pruning reservoir’s connections on an Echo State Network, based on the idea of significance of a connection. Our experiments show that we can obtain optimally sparse reservoirs, with a higher or comparable generalization accuracy. Read the full paper.
  2. GP-Based Kernel Evolution for L2-Regularization Networks: we detail a new GP-based procedure for evolving optimal kernels. We explore it using an L2-Regularization Network as base classifier. Moreover, we analyze the effect that diversity plays in the evolution process. (Full paper yet to appear)

I have also presented the paper An Interpretable Graph-Based Image Classifier (check it on IEEE Xplore). If you are interested in that presentation, please contact the corresponding author.