Non-volatile memory (NVM) crossbars have been identified as a promising
...
Modern datacenter applications are prone to high tail latencies since th...
Recent years have seen an increase in the development of large deep lear...
The rapid scaling of language models is motivating research using
low-bi...
In this paper, we propose TAPA, an end-to-end framework that compiles a ...
Graph neural networks (GNNs), which have emerged as an effective method ...
Pruning is a popular technique for reducing the model size and computati...
Hyperdimensional computing (HDC) is an emerging learning paradigm that
c...
Graph neural networks (GNNs) have been increasingly deployed in various
...
Top-1 ImageNet optimization promotes enormous networks that may be
impra...
As the vision of in-network computing becomes more mature, we see two
pa...
Neural network robustness has become a central topic in machine learning...
Depthwise separable convolutions and frequency-domain convolutions are t...
The ongoing shift of cloud services from monolithic designs to microserv...
Artificial intelligence (AI) technologies have dramatically advanced in
...
A black-box spectral method is introduced for evaluating the adversarial...
Binary neural networks (BNNs) have 1-bit weights and activations. Such
n...
This paper proposes GuardNN, a secure deep neural network (DNN) accelera...
Graph neural networks (GNNs) are gaining increasing popularity as a prom...
Cloud applications are increasingly relying on hundreds of loosely-coupl...
In this paper, we propose MgX, a near-zero overhead memory protection sc...
Field-programmable gate arrays (FPGAs) provide an opportunity to co-desi...
We propose precision gating (PG), an end-to-end trainable dynamic
dual-p...
Outliers in weights and activations pose a key challenge for fixed-point...
Graph embedding techniques have been increasingly deployed in a multitud...
Physical design process commonly consumes hours to days for large design...
Quantization can improve the execution latency and energy efficiency of
...
Quantization can improve the execution latency and energy efficiency of
...
We propose unitary group convolutions (UGConvs), a building block for CN...
Employing deep neural networks to obtain state-of-the-art performance on...
State-of-the-art convolutional neural networks are enormously costly in ...