Dileep's Vision Research Page

Welcome 

This page is intended to be of help to people who are interested in knowing more about research conducted according to the framework described in On Intelligence. I have been working with Jeff for over a year now and I have been assimmilating, formalizing and in some cases extending the ideas described in the book. 

Given the general goals and the assumptions described in the book, I concentrate on the visual system to figure out how the general principles play out to achieve the properties that human visual systems demonstrate. In particular, my current focus is on what is known as the Invariance Problem - how our visual system is able to recognize objects despite the different images that project on the retina. In pages that follow I will discuss the Invariance problem and describe my research on solving it the way brain does it.

Presently available are the following.

 NEW! Belief Propagation and Wiring Length Optimization as Organizing Principles for Cortical Microcircuits.
        This is  a speculative and unpublished manuscript on the connection between cortical microcircuits and Bayesian Belief Propagation. I think that the connection exists, but many of the details in this manuscript could be wrong. Still I would stand by the section on implications to understanding cortical anatomy.
 
 A Hierarchical Bayesian Model of Invariant Pattern Recognition in the Visual Cortex.
 Accepted for publication in the proceedings of  the International Joint Conference on Neural Networks. (IJCNN 05)

Invariant Pattern Recognition Using Bayesian Inference on Hierarchical Sequences 
This is a slightly older RNI technical report. It is available in pdf and html formats. 

    Table of contents for the upcoming webpage.
     

    NEW! I am co-organizing a workshop on Invariant Representations in Vision as part of the CoSyNe conference.
    Click here for the workshop website.


This page is under construction. I plan to make the following available as I make progress in my research.
    A description of the Invariance Problem and prior research. 
    A description of my formulation and approach. 
    A gallery of results achieved by my research. NOW AVAILABLE!!

    Source code for simulations (Version 1)

    Source code for simulations Version 2!! NOW AVAILABLE!!!! 

      This version incorporates feedback and was used to obtain the prediction results shown in the gallery.
    NIPS DEMO VERSION - NOW AVAILABLE!!!  (Dec 22 2004)
      I demonstrated this system at the Neural Information Processing Systems (NIPS) Conference. Click here to
      download this version.
    A description of future research I intend to undertake. 
     
Thank you for your interest in my research. 

Dileep George
Email. dgeorge {{at}} numenta {{dot}} com