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 language | English |
|---|---|
| Title of host publication | Proceedings - 2018 IEEE 27th Asian Test Symposium, ATS 2018 |
| Publisher | IEEE Computer Society |
| Pages | 173-178 |
| Number of pages | 6 |
| ISBN (Electronic) | 9781538694664 |
| DOIs | |
| Publication status | Published - 2018 Dec 6 |
| Event | 27th IEEE Asian Test Symposium, ATS 2018 - Hefei, China Duration: 2018 Oct 15 → 2018 Oct 18 |
Publication series
| Name | Proceedings of the Asian Test Symposium |
|---|---|
| Volume | 2018-October |
| ISSN (Print) | 1081-7735 |
Other
| Other | 27th IEEE Asian Test Symposium, ATS 2018 |
|---|---|
| Country | China |
| City | Hefei |
| Period | 18-10-15 → 18-10-18 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Electrical and Electronic Engineering
Cite this
}
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 proceeding › Conference 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 -