Jim Mutch Email:

I am a Ph.D. student in the department of Brain and Cognitive Sciences at MIT, where I work in the Center for Biological and Computational Learning. My interests include computational neuroscience and brain-inspired AI.

In 2006 I completed a masters degree at the University of British Columbia in Vancouver, Canada, where I was part of the Laboratory for Computational Intelligence in the department of Computer Science. Most of my masters research, supervised by David Lowe, was based on a biologically-inspired model of visual object recognition.

Papers

• Jim Mutch, Ulf Knoblich, and Tomaso Poggio. CNS: a GPU-based framework for simulating cortically-organized networks. MIT-CSAIL-TR-2010-013 / CBCL-286, Massachusetts Institute of Technology, Cambridge, MA, February 26, 2010. [pdf] [bib] [source code]

• Jim Mutch and David G. Lowe. Object class recognition and localization using sparse features with limited receptive fields. International Journal of Computer Vision (IJCV), 80(1), pp. 45-57, October 2008. [pdf] [bib] [source code]

• Jim Mutch and David G. Lowe. Multiclass Object Recognition with Sparse, Localized Features. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11-18, New York, June 2006. [pdf] [bib] [source code]

• Nando de Freitas, Richard Dearden, Frank Hutter, Ruben Morales-Menendez, Jim Mutch and David Poole. Diagnosis by a Waiter and a Mars Explorer. Proceedings of the IEEE Special Issue on Sequential State Estimation, March 2004. [pdf]

Code

• CNS: a general GPU-based framework for the fast simulation of "cortically-organized" networks, defined as networks consisting of n-dimensional layers of similar cells. This is a fairly broad class; we have also developed the following specialized CNS packages:
  • hmax: for building HMAX models (a reimplementation of FHLib in CNS).
  • cnpkg: 3-D convolutional networks with backprop.
  • hhpkg: networks of Hodgkin-Huxley spiking cells (written by Ulf Knoblich).

  (Note: while CNS is designed for use with a GPU, it can also run -- much more slowly -- without one.)

• hmin: a minimal implementation of HMAX, intended for illustration.

• FHLib: legacy code used in our CVPR 2006 and IJCV 2008 papers.