Co-simulation methodology for improved design and verification of hardware neural networks

Abstract

This paper presents a methodology for speeding up the design and verification of process of artificial neural networks (ANNs) in system-on-chip (SoC) hardware with the help of cosimulation. Application of advanced design methodologies for complex designs such as ANNs are important in todays fast moving hardware design industry. However, it is difficult to fully verify the functionality of ANNs when designed in hardware. Most forms of ANN require the use of complex training algorithms, which are difficult to implement in a testbench even with the help of modern interfaces such as SystemVerilog's DPI-C. The neural network topology selected as the case study for this paper are evolvable block-based neural networks (BbNNs). The case studies employed during the verification process are the XOR problem, driver drowsiness classification, and heart arrhythmia classification. The proposed methodology significantly reduces the verification time required for the design of hardware-based neural networks. This allows complex models applicable a variety of applications to be quickly designed, such as industrial motor controllers or fuzzy systems.

A Survey of Neuromorphic Computing and Neural Networks in Hardware

“…There are a variety of different and diverse application areas in this space to which neuromorphic systems have been applied, including accident diagnosis [1015], cereal grain identification [657], [659], computer user analysis [2641], [2642], driver drowsiness detection [2434], gas recognition or detection [622], [943], [972], product classification [781], hyperspectral data classification [750], stock price prediction [1537], wind velocity estimation [939], solder joint classification [1050], solar radiation detection [929], climate prediction [966], and applications within high energy physics [982], [1018]. Applications within the medical domain have also been popular, including coronary disease diagnosis [954], pulmonary disease diagnosis [887], deep brain sensor monitoring [404], DNA analysis [1146], heart arrhythmia detection [2434], analysis of electrocardiogram (ECG) [900], [922], [965], [1007], electroencephalogram (EEG) [403], [1100], [1103], [1850], and electromyogram (EMG) [1039], [1103] results, and pharmacology applications [2643]. A set of benchmarks from the UCI machine learning repository [2644] and/or the Proben1 data set [710], [908], [974], [2444] have been popular in both neural networ...…”

Target Recognition Based on Rough Set with Multi-Source Information

“…In essence, identification with multi-attribute means the process of integrating multi-source information. In the field of target recognition technology, the main mathematical analysis methods include The Bayes estimation [1], DS evidence theory [2], fuzzy set theory [3][4][5][6][7][8], neural networks [9][10][11][12][13][14][15], artificial intelligence technology [16], rough set theory [17][18][19][20][21], and so on.…”