Please wait ...
0% Complete
Home
/
12th International Conference on Computer and Knowledge Engineering
FAST: FPGA Acceleration of Neural Networks Training
Authors :
Alireza Borhani
1
Mohammad Hossein Goharinejad
2
Hamid Reza Zarandi
3
1- Department of Computer Engineering, Amirkabir university of technology
2- Department of Computer Engineering, Amirkabir university of technology
3- Department of Computer Engineering, Amirkabir university of technology
Keywords :
Field Programmable Gate Array،Embedded Devices،Artificial Neural Network،Machine Learning،Approximation
Abstract :
Training state-of-the-art ANNs is computationally and memory intensive. Thus, implementing the training on embedded devices with limited resources is challenging. In order to address this challenge, we propose FAST, a low-precision method to implement and optimize ANN training on FPGA. FAST first addresses the challenge of implementing the non-polynomial sigmoid activation function by presenting a solution using PNLA methods. Then, it introduces Hardware Optimized PReLU (HOPE) activation function, which is specifically devised to reduce the required resources and increase the accuracy of computations on FPGA. We evaluated FAST against the software implementations of ANNs, using training tasks available in the MNIST benchmark. The results show that FAST improves the training speed by 8.6× and reduces the required memory size by orders of magnitude. It is worthwhile to mention that the method imposes almost no degradation in training accuracy.
Papers List
List of archived papers
FAST: FPGA Acceleration of Neural Networks Training
Alireza Borhani - Mohammad Hossein Goharinejad - Hamid Reza Zarandi
Virtual Network Embedding based on Univariate Distribution Estimation
Arezoo Jahani
LPCNet: Lane detection by lane points correction network in challenging environments based on deep learning
Sina BaniasadAzad - Seyed Mohammadreza Mousavi mirkolaei
An effective hybrid algorithm for locating splicing forgery image
Seyed Hesamoddin Hosseini - Amene Vatanparast - Amir Hossein Taherinia
Ensemble-Based Fraud Detection: A Robust Approach Evaluated on IEEE-CIS
Fatemeh Moradi - Mehran Tarif - Mohammadhossein Homaei
A Genetic-based Fusion Approach of Persian and Universal Phonetic results for Spoken Language Identification
Ashkan Moradi - Yasser Shekofteh - Saeed Zarei
Implementation of a Low-Overhead 2-Bit Parity-Preserving Reversible Vedic Multiplier for Quantum Architectures
Shekoofeh Moghimi - Negin Mashayekhi - Mohammad Reza Reshadinezhad
Capsule Routing over Stacked GCN-GAT Embeddings with Negative Sampling for Graph Link Prediction
Fatemeh Safari Sarvandi - Sayeh Mirzaei - Rooholah Abedian
Prediction of rTMS Treatment Response in Depression Using a Frequency-Based EEG Biomarker
Ali Asadi Zeidabadi - Saeid Rashidi
Semantic Segmentation Using Region Proposals and Weakly-Supervised Learning
Maryam Taghizadeh - Abdolah Chalechale
more
Samin Hamayesh - Version 46.0.2