Software and Projects

 

Sparse HMAX

HMAX is a biology inspired model developed in Poggio's lab at MIT during 1999 to 2007. We substitute its learning methods with spase coding or ICA and termed the resulted model sparse HMAX. MATLAB codes are are available.

Version 1.1 (download)

This is the version we used to generate figures in (Hu, et al., 2014). Only ICA is available for learning bases. If you execute the codes you'll see three figures similar to the following. From left to right: S1 bases (size 10x10), S2 bases (size 45x45), S3 bases (size 120x120).

 

Version 1.2 (download)

Sparse coding is provided as an optional algorithm for learning bases. If you choose sparse coding, you'll see three figures similar to the following. From left to right: S1 bases (size 10x10), S2 bases (size 45x45), S3 bases (size 120x120). If you choose ICA, the results will be a little bit different from above as I have modified the original pipeline to accommodate both ICA and sparse coding.