51 citations · 138 across the 8 of their papers we have counts for
9 papers
PIM-GPT: A Hybrid Process-in-Memory Accelerator for Autoregressive Transformers
Yuting Wu, Ziyu Wang, Wei D. Lu
Decoder-only Transformer models such as GPT have demonstrated exceptional performance in text generation, by autoregressively predicting the next token. However, the efficacy of ru…
Bulk-Switching Memristor-based Compute-In-Memory Module for Deep Neural Network Training
Yuting Wu, Qiwen Wang, Ziyu Wang +5
The need for deep neural network (DNN) models with higher performance and better functionality leads to the proliferation of very large models. Model training, however, requires in…
RN-Net: Reservoir Nodes-Enabled Neuromorphic Vision Sensing Network
Sangmin Yoo, Eric Yeu-Jer Lee, Ziyu Wang +2
Event-based cameras are inspired by the sparse and asynchronous spike representation of the biological visual system. However, processing the event data requires either using expen…
Side-channel attack analysis on in-memory computing architectures
Ziyu Wang, Fan-hsuan Meng, Yongmo Park +2
In-memory computing (IMC) systems have great potential for accelerating data-intensive tasks such as deep neural networks (DNNs). As DNN models are generally highly proprietary, th…
Gradient-based Neuromorphic Learning on Dynamical RRAM Arrays
Peng Zhou, Jason K. Eshraghian, Dong-Uk Choi +2
We present MEMprop, the adoption of gradient-based learning to train fully memristive spiking neural networks (MSNNs). Our approach harnesses intrinsic device dynamics to trigger n…
Navigating Local Minima in Quantized Spiking Neural Networks
Jason K. Eshraghian, Corey Lammie, Mostafa Rahimi Azghadi +1
Spiking and Quantized Neural Networks (NNs) are becoming exceedingly important for hyper-efficient implementations of Deep Learning (DL) algorithms. However, these networks face ch…
The fine line between dead neurons and sparsity in binarized spiking neural networks
Jason K. Eshraghian, Wei D. Lu
Spiking neural networks can compensate for quantization error by encoding information either in the temporal domain, or by processing discretized quantities in hidden states of hig…
Design Space Exploration of Dense and Sparse Mapping Schemes for RRAM Architectures
Corey Lammie, Jason K. Eshraghian, Chenqi Li +4
The impact of device and circuit-level effects in mixed-signal Resistive Random Access Memory (RRAM) accelerators typically manifest as performance degradation of Deep Learning (DL…
Hierarchical Architectures in Reservoir Computing Systems
John Moon, Wei D. Lu
Reservoir computing (RC) offers efficient temporal data processing with a low training cost by separating recurrent neural networks into a fixed network with recurrent connections…