This Article 
 Bibliographic References 
 Add to: 
High Performance Fault-Tolerant Digital Neural Networks
March 1998 (vol. 47 no. 3)
pp. 357-363

Abstract—Efficient implementation of neural networks requires high-performance architectures, while VLSI realization for mission-critical applications must include fault tolerance. Contemporaneous solution of such problems has not yet been completely afforded in the literature. This paper focuses both on data representation to support high-performance neural computation and on error detection to provide the basic information for fault tolerance by using the redundant binary representation with a three-rail logic implementation. Costs and performances are evaluated referring to multilayered feed-forward networks.

[1] J. Hertz and A. Krogh, Introduction to the Theory of Neural Computation. Wesley Publishing Co., 1991.
[2] C. Lehmann and F. Blayo, "A VLSI Implementation of a Generic Systolic Synaptic Building Block for Neural Networks," Proc. Int'l Workshop VLSI for Artificial Intelligence and Neural Networks, pp. G7.1-10,Oxford, U.K., Sept. 1990.
[3] S.Y. Kung and J.N. Hwang, "A Unified Systolic Architecture for Artificial Neural Networks," Proc. Int'l Conf. Systolic Arrays, pp. 163-170,San Diego, Calif., 1988.
[4] V. Piuri, M. Sami, and R. Stefanelli, "Fault Tolerance in Neural Networks: Theoretical Analysis and Simulation Results," IEEE Proc. Compeuro 1991, pp. 429-436,Bologna, Italy, May 1991.
[5] M. Stevenson, R. Winter, and B. Widrow, "Sensitivity of Feedforward Neural Networks to Weights Errors," IEEE Trans. Neural Networks, vol. 1, no 1, Mar. 1990.
[6] C. Alippi, V. Piuri, and M. Sami, "Sensitivity to Errors in Artificial Neural Networks: A Behavioural Approach," IEEE Trans. Circuits and Systems-I, vol. 42, no 6, June 1995.
[7] L. Breveglieri and V. Piuri, "Error Detection in Digital Neural Networks: An Algorithm-Based Approach for Inner Product Protection," Proc. 1994 SPIE Conf.—Advance Signal Processing, pp. 809-820,San Diego, Calif., July 1994.
[8] Y.-M. Hsu, V. Piuri, and E.E. Swartzlander Jr., "Time-Redundant Multiple Computation for Fault-Tolerant Digital Neural Networks," 1995 IEEE Proc. Int'l Symp. Circuits and Systems, pp.II.977-II.980,Seattle, Wash., Apr. 1995.
[9] Y.-M. Hsu, V. Piuri, and E.E. Swartzlander Jr., "Recomputing by Operand Exchanging: A Time Redundancy Approach for Fault-Tolerant Neural Networks," 1995 IEEE Proc. Int'l Conf. Application-Specific Array Processors, pp. 54-65,Strasbourg, France, July 1995.
[10] V. Piuri, M. Sami, and R. Stefanelli, "Arithmetic Codes for Concurrent Error Detection in Artificial Neural Networks: The Case of AN+B Codes," IEEE Proc. Int'l Workshop Defect and Fault Tolerance in VLSI Systems, pp. 277-286,Dallas, Tex., Nov. 1992.
[11] V. Piuri and M. Villa, "Residue Codes for Concurrent Error Detection in Artificial Neural Networks," Proc. IJCNN '93, pp. IV.825-IV.830,Portland, Ore., July 1993.
[12] A. Avizenis, "Signed-Digit Number Representations for Fast Parallel Arithmetic," IRE Trans. Electronic Computers, vol. 10, pp. 389-400, Sept. 1961.
[13] J.E. Smith and G. Metze, "Strongly Fault Secure Logic Networks," IEEE Trans. Computers, vol. 27, pp. 491-499, June 1978.
[14] N. Takagi and S. Yajima, "On-Line Error-Detectable High-Speed Multiplier Using Redundant Binary Representation and Three-Rail Logic," IEEE Trans. Computers, vol. 36, no. 11, pp. 1,310-1,317, Nov. 1987.
[15] T.R.N. Rao, Error Coding for Arithmetic Processors.New York: Academic Press, 1974.
[16] R. David, "Totally Self-Checking 1-out-of-3 Code Checker," IEEE Trans. Computers, vol. 27, pp. 570-572, June 1978.
[17] P. Golan, "Design of Totally Self-Checking Checker for 1-out-of-3 Code," IEEE Trans. Computers, vol. 33, no. 3, p. 285, Mar. 1984.
[18] A.M. Paschalis, C. Efstathiou, and C. Halatsis, "An Efficient TSC 1-out-of-3 Code Checker," IEEE Trans. Computers, vol. 39, no. 3, pp. 407-411, Mar. 1990.
[19] H. Hao and E.J. McCluskey, "Resistive Shorts within CMOS gates," Proc. IEEE Int'l Test Conf., pp. 292-301, 1991.
[20] C. Metra, M. Favalli, P. Olivo, and B. Riccò, "A Highly Testable 1-out-of 3 CMOS Checker," Proc. IEEE Int'l Workshop Defect and Fault Tolerance in VLSI Systems, pp. 279-286,Venice, Italy, Oct. 1993.
[21] A.M. Usas, "A Totally Self-Checker Design for the Detection of Errors in Periodic Signals," IEEE Trans. Computers, vol. 24, pp. 483-488, May 1975.

Index Terms:
Neural architecture, redundant binary representation, three-rail logic, concurrent error detection, unidirectional errors, high-performance architecture.
Simone Bettola, Vincenzo Piuri, "High Performance Fault-Tolerant Digital Neural Networks," IEEE Transactions on Computers, vol. 47, no. 3, pp. 357-363, March 1998, doi:10.1109/12.660173
Usage of this product signifies your acceptance of the Terms of Use.