On the Reliability of Linear Regression and Pattern Recognition Feedforward Artificial Neural Networks in FPGAs

In this paper, we experimentally and analytically evaluate the reliability of two state-of-the-art neural networks for linear regression and pattern recognition (multilayer perceptron and single-layer perceptron) implemented in a system-on-chip composed of a field-programmable gate array (FPGA) and a microprocessor. We have considered, for each neural network, three different activation function complexities, to evaluate how the implementation affects FPGAs reliability. As we show in this paper, higher complexity increases the exposed area but reduces the probability of one failure to impact the network output. In addition, we propose to distinguish between critical and tolerable errors in artificial neural networks. Experiments using a controlled heavy-ions beam show that, for both networks, only about 30% of the observed output errors actually affect the outputs correctness. We identify the causes of critical errors through fault injection, and found that faults in initial layers are more likely to significantly affect the output.

[1]  Michael J. Pazzani,et al.  A Principal Components Approach to Combining Regression Estimates , 1999, Machine Learning.

[2]  Eduardo Chielle,et al.  Analyzing the Impact of Radiation-Induced Failures in Programmable SoCs , 2016, IEEE Transactions on Nuclear Science.

[3]  Heather Quinn,et al.  A Method and Case Study on Identifying Physically Adjacent Multiple-Cell Upsets Using 28-nm, Interleaved and SECDED-Protected Arrays , 2014, IEEE Transactions on Nuclear Science.

[4]  R. Fisher.  THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .

[5]  Michael J. Wirthlin,et al.  High-Reliability FPGA-Based Systems: Space, High-Energy Physics, and Beyond , 2015, Proceedings of the IEEE.

[6]  F ROSENBLATT,et al.  The perceptron: a probabilistic model for information storage and organization in the brain. , 1958, Psychological review.

[7]  R. Castano,et al.  Creation and testing of an artificial neural network based carbonate detector for Mars rovers , 2005, 2005 IEEE Aerospace Conference.

[8]  P. Werbos,et al.  Beyond Regression : "New Tools for Prediction and Analysis in the Behavioral Sciences , 1974 .

[9]  Phan Quoc Dzung,et al.  The new MPPT algorithm using ANN-based PV , 2010, International Forum on Strategic Technology 2010.

[10]  Naser Pariz,et al.  FPGA implementation of an ANN for detection of anthelmintics resistant nematodes in sheep flocks , 2009, 2009 4th IEEE Conference on Industrial Electronics and Applications.

[11]  Victor-Emil Neagoe,et al.  A concurrent neural network approach to pedestrian detection in thermal imagery , 2012, 2012 9th International Conference on Communications (COMM).

[12]  K. M. Curtis,et al.  Piecewise linear approximation applied to nonlinear function of a neural network , 1997 .

[13]  Christopher C. Doss,et al.  FPGA-based implementation of a robust IEEE-754 exponential unit , 2004, 12th Annual IEEE Symposium on Field-Programmable Custom Computing Machines.

[14]  S. Tahar,et al.  Design and synthesis of an IEEE-754 exponential function , 1999, Engineering Solutions for the Next Millennium. 1999 IEEE Canadian Conference on Electrical and Computer Engineering (Cat. No.99TH8411).

[15]  Deming Chen,et al.  A novel SoC architecture on FPGA for ultra fast face detection , 2009, 2009 IEEE International Conference on Computer Design.

[16]  L. E. Seixas,et al.  Experimental setup for Single Event Effects at the São Paulo 8UD Pelletron Accelerator , 2014 .

[17]  W. Pitts,et al.  A Logical Calculus of the Ideas Immanent in Nervous Activity (1943) , 2021, Ideas That Created the Future.

[18]  D. Rubinfeld,et al.  Hedonic housing prices and the demand for clean air , 1978 .

[19]  Rizwan Beg,et al.  Genetic neural network based data mining in prediction of heart disease using risk factors , 2013, 2013 IEEE CONFERENCE ON INFORMATION AND COMMUNICATION TECHNOLOGIES.

[20]  Fernanda Gusmão de Lima Kastensmidt,et al.  Method to Analyze the Susceptibility of HLS Designs in SRAM-Based FPGAs Under Soft Errors , 2016, ARC.

[21]  Gabriel L. Nazar,et al.  Reducing embedded software radiation-induced failures through cache memories , 2014, 2014 19th IEEE European Test Symposium (ETS).

[22]  Prapan Leekul,et al.  Low complexity artificial neural network unit for sugar content detection in microwave sensor system , 2014, Signal and Information Processing Association Annual Summit and Conference (APSIPA), 2014 Asia-Pacific.