The Correspondence Between Deterministic and Stochastic Digital Neurons: Analysis and Methodology

• Published in: IEEE transactions on neural networks Vol. 19; no. 10; pp. 1739 - 1752
• Main Authors: Geretti, L., Abramo, A.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.10.2008; Institute of Electrical and Electronics Engineers
• Subjects: Algorithms; Applied sciences; Availability; Computer Simulation; Deterministic equivalence; Electronics; Exact sciences and technology; Feedback; Field programmable gate arrays; Filling; FPGA implementation; Guidelines; Integrated circuits; Integrated circuits by function (including memories and processors); Models, Statistical; Neural network hardware; Neural networks; Neural Networks (Computer); Neurons; Numerical Analysis, Computer-Assisted; Resource management; Semiconductor electronics. Microelectronics. Optoelectronics. Solid state devices; Signal Processing, Computer-Assisted; stochastic neuron; Stochastic Processes; Stochastic resonance; Stochastic analysis; Statistical analysis; Probabilistic approach; Circuit design; Activation function; Bias; Deterministic equivalence; FPGA implementation; Field programmable gate array; Neural network; Deterministic approach; stochastic neuron
• ISSN: 1045-9227; 1941-0093
• DOI: 10.1109/TNN.2008.2001775

This paper analyzes the criteria for the direct correspondence between a deterministic neural network and its stochastic counterpart, and presents the guidelines that have been derived to establish such a correspondence during the design of a neural network application. In particular, the role of the slope and bias of the neuron activation function and that of the noise of its output have been addressed, thus filling a specific literature gap. This paper presents the results that have been theoretically derived in this regard, together with the simulations of few relevant application examples that have been performed to support them.