Abstract:Resistive random access memory (RRAM) is a promising technology for energy-efficient in-memory computing. However, due to technology limits, RRAM device faces a series of reliability issues. Deep neural network (DNN) computing based on RRAM suffers from accuracy degradation. On the one hand, offline DNN training solutions are difficult to fully consider and simulate all nonidealities. Worse still, new error or nonideality may come up with the usage of RRAM, which further deteriorates the effectiveness of offline training. On the other hand, online training poses great challenges on programming overhead and device lifetime. The iterative writeverify technique to program multi-bit RRAM cells prolongs write latency more than 10x longer than read latency. To overcome these issues, we propose a compensation architecture and a software and hardware co-training design to mitigate the realistic network accuracy loss in RRAM-based computing-inmemory chips. Firstly, we add trainable compensation channels in crossbars utilizing the residual resource after original weight mapping. Secondly, an offline training procedure with computing output from hardware is triggered to settle down appropriate weight value in compensation channels. Experimental results demonstrate that the proposed design can guarantee <= 0.8% loss of accuracy in DNN on MNIST and CIFAR10 dataset even when nonidealities reduce the original accuracy down to =73%.

Impacts of State Instability and Retention Failure of Filamentary Analog RRAM on the Performance of Deep Neural Network

Analytic Model for Statistical State Instability and Retention Behaviors of Filamentary Analog RRAM Array and Its Applications in Design of Neural Network

Robustness of Neuromorphic Computing with RRAM-based Crossbars and Optical Neural Networks.

Improvement of State Stability in Multi-Level Resistive Random-Access Memory (RRAM) Array for Neuromorphic Computing

Oxide-based filamentary RRAM for deep learning

Early-Stage Fluctuation in Low-Power Analog Resistive Memory: Impacts on Neural Network and Mitigation Approach

Investigation of Statistical Retention of Filamentary Analog RRAM for Neuromophic Computing

Compensation Architecture to Alleviate Noise Effects in RRAM-based Computing-in-memory Chips with Residual Resource

An RRAM Retention Prediction Framework Using a Convolutional Neural Network Based on Relaxation Behavior

The Impact of Endurance Degradation in Analog RRAM for In-Situ Training

Lifetime Improvement of 28 Nm Resistive Random Access Memory Chip by Machine Learning‐Assisted Prediction Model Collaborated with Resurrection Algorithm

Circuit Modeling for RRAM-Based Neuromorphic Chip Crossbar Array with and Without Write-Verify Scheme.

A Study of the Effect of RRAM Reliability Soft Errors on the Performance of RRAM-Based Neuromorphic Systems

Thermal Induced Retention Degradation of RRAM-based Neuromorphic Computing Chips

Compensation Architecture Design Utilizing Residual Resource to Mitigate Impacts of Nonidealities in RRAM-based Computing-in-memory Chips

Bit-Aware Fault-Tolerant Hybrid Retraining and Remapping Schemes for RRAM-Based Computing-in-Memory Systems

Statistical Temperature Coefficient Distribution in Analog RRAM Array: Impact on Neuromorphic System and Mitigation Method

Exploring the Impact of Random Telegraph Noise-Induced Accuracy Loss on Resistive RAM-Based Deep Neural Network

Drop-Connect as a Fault-Tolerance Approach for RRAM-based Deep Neural Network Accelerators

An On-chip Layer-wise Training Method for RRAM Based Computing-in-memory Chips.

Endurance and Retention Degradation of Intermediate Levels in Filamentary Analog RRAM