Electrically modifiable nonvolatile sonos synapses for electronic neural networks. Semiannual technical report, 1 Mar-1 Sep 91

Technical Report · Sun Sep 01 00:00:00 EDT 1991

OSTI ID:6132894

White, M H; Chen, C Y.M.; French, M

The current surge of enthusiasm for neural network aims to construct systems that can learn or modify their behavior according to the environment. There are many similarities which exist between this new class of machine and human beings. One of these similarities is the massive parallelism in processing information. Parallel processing concepts are in stark contrast to the operations of modern digital computers that perform large numbers of sequential operations very rapidly and accurately. Researchers believe the synaptic junctions in a neutral system are the local memory sites and provide the physiological basis for the distributed parallel systems. These synapses are not only modifiable but also serve the functions of storing and transmitting information from neuron to neuron. To reduce the complex modelling required for the synaptic interconnection, the representation of the synapse has been simplified to a single ideal junction between the output of neurons (axons) and the inputs to neurons (dendrites). Synaptic modifications requires information from the input and the output of the neuron in order to perform complex recognition. Therefore, the nature of the synaptic junction and the principle or algorithm which controls local organization at the neuron level become two central issues pertaining to neural networks research.

• Research Organization: Lehigh Univ., Bethlehem, PA (United States). Sherman Fairchild Center for Solid State Studies
• OSTI ID: 6132894
• Report Number(s): AD-A-242213/7/XAB; CNN: N00014-89-J-3149
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
NEURAL NETWORKS
RESEARCH PROGRAMS
ALGORITHMS
CONTROL
DATA PROCESSING
DIGITAL COMPUTERS
DISTRIBUTED DATA PROCESSING
DISTRIBUTION
ELECTRONIC EQUIPMENT
JUNCTIONS
LEARNING
MEMORY DEVICES
MODIFICATIONS
NERVE CELLS
PARALLEL PROCESSING
SURGES
ANIMAL CELLS
COMPUTERS
EQUIPMENT
MATHEMATICAL LOGIC
PROCESSING
PROGRAMMING
SOMATIC CELLS
990200* - Mathematics & Computers