Performance Optimization for Pattern Recognition Using Associative Neural Memory
Abstract
This paper describes the performance optimization in software and hardware solutions for a cognitive computing model called Brain State in a Box (BSB). This BSB model is implemented using two different configurations of the proposed architecture. The first implementation is a software only approach using the Cell Broadband Engine. The other implementation is a hybrid configurable computing platform which uses Field Programmable Gate Array (FPGA) for implementing the computation. To compensate its efficiency, the BSB based associative neural memory is applied for symbol and character recognition.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2008
- Accession Number
- ADA502126
Entities
People
- Daniel Burns
- Michael J Moore
- Prakash Mukre
- Qing Wu
- Qinru Qiu
- Richard Linderman
- Tom Renz
Organizations
- Air Force Research Laboratory