Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM

Tsung-Chu Huang, Jeffae Schroff

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Neuromorphic RRAM has become the most promising candidate for AI applications. But it suffers three issues including degradation, defects and errors. To overcome the three issues we proposed a precompensation technique for compensating resistive degradation. A linear-system-based BIST architecture is developed with proposed diagonal sliding march test can effectively and efficiently screen out the uncompensated degradation and permanent defects. Analog Berger codes is proposed for detecting transient errors for variation learning and self-checking for asymmetric errors. From evaluations, the precompensation takes only 5/B time for batch operations of B cycles. Proposed BIST approach and method can reduced 2LN march tests to 6N for L-level RRAMs. The self-healing ability is verified by analog-Berger-code error detection. From experiments using a typical neural network for MNIST handwritten digit dataset the network can be healed with only 2% of accuracy and about 35% of training steps.

Original languageEnglish
Title of host publicationProceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018
PublisherIEEE Computer Society
Pages173-178
Number of pages6
ISBN (Electronic)9781538694664
DOIs
Publication statusPublished - 2018 Dec 6
Event27th IEEE Asian Test Symposium, ATS 2018 - Hefei, China
Duration: 2018 Oct 152018 Oct 18

Publication series

NameProceedings of the Asian Test Symposium
Volume2018-October
ISSN (Print)1081-7735

Other

Other27th IEEE Asian Test Symposium, ATS 2018
CountryChina
CityHefei
Period18-10-1518-10-18

Fingerprint

Built-in self test
Degradation
Defects
Error detection
Linear systems
Neural networks
RRAM
Experiments

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

Huang, T-C., & Schroff, J. (2018). Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM. In Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018 (pp. 173-178). [8567430] (Proceedings of the Asian Test Symposium; Vol. 2018-October). IEEE Computer Society. https://doi.org/10.1109/ATS.2018.00041
Huang, Tsung-Chu ; Schroff, Jeffae. / Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM. Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018. IEEE Computer Society, 2018. pp. 173-178 (Proceedings of the Asian Test Symposium).
@inproceedings{55ee461b8b2d48bb92b90fd101deeac8,
title = "Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM",
abstract = "Neuromorphic RRAM has become the most promising candidate for AI applications. But it suffers three issues including degradation, defects and errors. To overcome the three issues we proposed a precompensation technique for compensating resistive degradation. A linear-system-based BIST architecture is developed with proposed diagonal sliding march test can effectively and efficiently screen out the uncompensated degradation and permanent defects. Analog Berger codes is proposed for detecting transient errors for variation learning and self-checking for asymmetric errors. From evaluations, the precompensation takes only 5/B time for batch operations of B cycles. Proposed BIST approach and method can reduced 2LN march tests to 6N for L-level RRAMs. The self-healing ability is verified by analog-Berger-code error detection. From experiments using a typical neural network for MNIST handwritten digit dataset the network can be healed with only 2{\%} of accuracy and about 35{\%} of training steps.",
author = "Tsung-Chu Huang and Jeffae Schroff",
year = "2018",
month = "12",
day = "6",
doi = "10.1109/ATS.2018.00041",
language = "English",
series = "Proceedings of the Asian Test Symposium",
publisher = "IEEE Computer Society",
pages = "173--178",
booktitle = "Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018",
address = "United States",

}

Huang, T-C & Schroff, J 2018, Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM. in Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018., 8567430, Proceedings of the Asian Test Symposium, vol. 2018-October, IEEE Computer Society, pp. 173-178, 27th IEEE Asian Test Symposium, ATS 2018, Hefei, China, 18-10-15. https://doi.org/10.1109/ATS.2018.00041

Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM. / Huang, Tsung-Chu; Schroff, Jeffae.

Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018. IEEE Computer Society, 2018. p. 173-178 8567430 (Proceedings of the Asian Test Symposium; Vol. 2018-October).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM

AU - Huang, Tsung-Chu

AU - Schroff, Jeffae

PY - 2018/12/6

Y1 - 2018/12/6

N2 - Neuromorphic RRAM has become the most promising candidate for AI applications. But it suffers three issues including degradation, defects and errors. To overcome the three issues we proposed a precompensation technique for compensating resistive degradation. A linear-system-based BIST architecture is developed with proposed diagonal sliding march test can effectively and efficiently screen out the uncompensated degradation and permanent defects. Analog Berger codes is proposed for detecting transient errors for variation learning and self-checking for asymmetric errors. From evaluations, the precompensation takes only 5/B time for batch operations of B cycles. Proposed BIST approach and method can reduced 2LN march tests to 6N for L-level RRAMs. The self-healing ability is verified by analog-Berger-code error detection. From experiments using a typical neural network for MNIST handwritten digit dataset the network can be healed with only 2% of accuracy and about 35% of training steps.

AB - Neuromorphic RRAM has become the most promising candidate for AI applications. But it suffers three issues including degradation, defects and errors. To overcome the three issues we proposed a precompensation technique for compensating resistive degradation. A linear-system-based BIST architecture is developed with proposed diagonal sliding march test can effectively and efficiently screen out the uncompensated degradation and permanent defects. Analog Berger codes is proposed for detecting transient errors for variation learning and self-checking for asymmetric errors. From evaluations, the precompensation takes only 5/B time for batch operations of B cycles. Proposed BIST approach and method can reduced 2LN march tests to 6N for L-level RRAMs. The self-healing ability is verified by analog-Berger-code error detection. From experiments using a typical neural network for MNIST handwritten digit dataset the network can be healed with only 2% of accuracy and about 35% of training steps.

UR - http://www.scopus.com/inward/record.url?scp=85060012892&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060012892&partnerID=8YFLogxK

U2 - 10.1109/ATS.2018.00041

DO - 10.1109/ATS.2018.00041

M3 - Conference contribution

AN - SCOPUS:85060012892

T3 - Proceedings of the Asian Test Symposium

SP - 173

EP - 178

BT - Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018

PB - IEEE Computer Society

ER -

Huang T-C, Schroff J. Precompensation, BIST and Analogue Berger Codes for Self-Healing of Neuromorphic RRAM. In Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018. IEEE Computer Society. 2018. p. 173-178. 8567430. (Proceedings of the Asian Test Symposium). https://doi.org/10.1109/ATS.2018.00041