Heatvit: Hardware-efficient adaptive token pruning for vision transformers

P Dong, M Sun, A Lu, Y Xie, K Liu… - … Symposium on High …, 2023 - ieeexplore.ieee.org
While vision transformers (ViTs) have continuously achieved new milestones in the field of
computer vision, their sophisticated network architectures with high computation and …

Order-of-magnitude differences in computational performance of analog Ising machines induced by the choice of nonlinearity

F Böhm, TV Vaerenbergh, G Verschaffelt… - Communications …, 2021 - nature.com
Ising machines based on nonlinear analog systems are a promising method to accelerate
computation of NP-hard optimization problems. Yet, their analog nature is also causing …

Stochastic configuration machines: FPGA implementation

MJ Felicetti, D Wang - arXiv preprint arXiv:2310.19225, 2023 - arxiv.org
Neural networks for industrial applications generally have additional constraints such as
response speed, memory size and power usage. Randomized learners can address some …

PLAC: Piecewise linear approximation computation for all nonlinear unary functions

H Dong, M Wang, Y Luo, M Zheng, M An… - … Transactions on Very …, 2020 - ieeexplore.ieee.org
This article presents a piecewise linear approximation computation (PLAC) method for all
nonlinear unary functions, which is an enhanced universal and error-flattened piecewise …

An efficient fpga-based convolutional neural network for classification: Ad-mobilenet

S Bouguezzi, HB Fredj, T Belabed, C Valderrama… - Electronics, 2021 - mdpi.com
Convolutional Neural Networks (CNN) continue to dominate research in the area of
hardware acceleration using Field Programmable Gate Arrays (FPGA), proving its …

A modular approximation methodology for efficient fixed-point hardware implementation of the sigmoid function

Z Pan, Z Gu, X Jiang, G Zhu… - IEEE Transactions on …, 2022 - ieeexplore.ieee.org
The sigmoid function is a widely used nonlinear activation function in neural networks. In this
article, we present a modular approximation methodology for efficient fixed-point hardware …

[HTML][HTML] Data multiplexed and hardware reused architecture for deep neural network accelerator

G Raut, A Biasizzo, N Dhakad, N Gupta, G Papa… - Neurocomputing, 2022 - Elsevier
Despite many decades of research on high-performance Deep Neural Network (DNN)
accelerators, their massive computational demand still requires resource-efficient, optimized …

Pancreas segmentation using a dual-input v-mesh network

Y Wang, G Gong, D Kong, Q Li, J Dai, H Zhang… - Medical Image …, 2021 - Elsevier
Accurate segmentation of the pancreas from abdomen scans is crucial for the diagnosis and
treatment of pancreatic diseases. However, the pancreas is a small, soft and elastic …

Energy efficient deep learning inference embedded on FPGA for sleep apnea detection

O Hassan, T Paul, MH Shuvo, D Parvin… - Journal of Signal …, 2022 - Springer
Sleep apnea is a type of disorder caused by the absence of breathing for a specific period of
time coupled with a significant decrease in the blood oxygen saturation level. The …

Fast and accurate approximation methods for trigonometric and arctangent calculations for low-performance computers

T Kusaka, T Tanaka - Electronics, 2022 - mdpi.com
In modern computers, complicated signal processing is highly optimized with the use of
compilers and high-speed processing using floating-point units (FPUs); therefore …