Title :
CHIMERA: creating a new generation chip by brain guidance
Author_Institution :
Dept. of Math., King´´s Coll., London, WC, UK
Abstract :
We present an outline for an adaptive chip architecture, based on general principles of global brain processing. The chip composed of sets of single neurons, which are initially grouped into columns, and these further grouped into modules. The overall architecture of the chip is based on the attention control system of the brain, together with its division into posterior low-level processing and classifying modules (of SOM type) and high-level cognitive modules (of recurrent type). Various balances and neuron complexity is discussed, as further aspects of the processing styles to be used (causal learning laws, salience maps and emotionality, working memory structures, inhibitory competition, neuromodulators).
Keywords :
adaptive systems; brain; learning (artificial intelligence); neural chips; recurrent neural nets; self-organising feature maps; CHIMERA chip; adaptive chip architecture; brain guidance; control system; global brain processing; high level cognitive modules; learning laws; neuromodulators; neuron complexity; posterior low level processing; recurrent neural net; self organising map; Adaptive control; Concurrent computing; Control systems; Humans; Intelligent robots; Neurons; Pressure control; Real time systems; Speech; Synchronization;
Conference_Titel :
Neural Networks, 2004. Proceedings. 2004 IEEE International Joint Conference on
Print_ISBN :
0-7803-8359-1
DOI :
10.1109/IJCNN.2004.1380920