Welcome | Projects | Publications | Technical Reports | Software | Resources | Sponsors | Personnel

Perceptual Engineering Research: Holographic Memory and Recollection

Keywords: Pattern Recognition, Target Recognition, Attention, Focus, Memory, Associative Recollection


In this research we are investigating the mathematical basis for complex attention modulated fast pattern recognition capability of naturally occuring systems beyond neural networks and potential means for highly accurate and speedy pattern recongnition.

Neural associative computing of 60-80's demonstrated few advantages extremely attractive to high performance real time pattern matching applications such as:

  • They are adaptive.
  • Arbitrary complex patterns can be learned through general purpose learning algorithms.
  • They are fast.  Retrieval operations are generally constant time operation. As a result, once learned, massive amount of patterns can be searched and retrieved in near real time.
  • They are also highly parallelizable.

Despite their superb characteristics, however, neural network based AAMs have found limited success in general pattern matching, except as a full frameadaptive filter, over the last 50 years. 

It seems still there are many features of natural memory which has not been grasped well. One such feature seems to be how do human handle dynamically changeable attention. In this perceptual engineering research we are searching computational models which might be able to recreate some of these attention phenomena. 

We show a particular new computational model called multidimensional holographic associative computing (MHAC) which unlike any existing neural network (NN) based artificial associative memories (AAM), can localize (or focus) its search on any subset of the pattern space, dynamically during retrieval.


 

Page last updated January 27, 2000, Networking and Media Communications Research Lab, Kent State University