Real-Time Neural Network Inversion on the SRC-6e Reconfigurable Computer

Abstract

Abstract-Implementation of real-time neural network inversion on the SRC-6e, a computer that uses multiple field-programmable gate arrays (FPGAs) as reconfigurable computing elements, is examined using a sonar application as a specific case study. A feedforward multilayer perceptron neural network is used to estimate the performance of the sonar system (Jung et al., 2001). A particle swarm algorithm uses the trained network to perform a search for the control parameters required to optimize the output performance of the sonar system in the presence of imposed environmental constraints (Fox et al., 2002). The particle swarm optimization (PSO) requires repetitive queries of the neural network. Alternatives for implementing neural networks and particle swarm algorithms in reconfigurable hardware are contrasted. The final implementation provides nearly two orders of magnitude of speed increase over a state-of-the-art personal computer (PC), providing a real-time solution.Index Terms-Field-programmable gate arrays (FPGAs), inverse problems, neural network hardware, particle swarm theory, real-time systems, reconfigurable architectures, sonar.

Evolutionary Inversion of Swarm Emergence Using Disjunctive Combs Control

“…Using Combs control, the inverse problem [11], which has found use in many areas of computational intelligence [33], [46], [47], [51], [52], [80], [81], reduces to adapting the actuator functions to maximize the fitness of the emergent behavior C. This is a variation of fuzzy membership function adaptation applied in Mamdani-type inference systems [6]- [8], [65], [88] and similar to inversion of trained neural networks. Although Mamdani control requires exponentially more tunable parameters than Combs control, their performances have been shown to be commensurate in many applications [22], [23], [94]- [97].…”

Parallel scalable hardware implementation of asynchronous discrete particle swarm optimization

“…In Duren et al (2007), the authors have used the PSO algorithm implemented on an SRC-6e reconfigurable computer to invert the neural network to search for a set of inputs to the network. The network is implemented in one FPGA and a 3-stage pipelined PSO is implemented in a separate Virtex-2 FPGA running at 100 MHz.…”

A Methodology for Implementation of the Execution Phase of Artificial Neural Networks in Digital Hardware Devices

“…(1) The Sigmoid function may be implemented by using Coordinate Rotation Digital Computer (CORDIC) algorithm [9], or by means of LUT [5] [10]. LUT is a simpler option with less area cost and processing time.…”