• Home
  • Search
  • ReLAccS: A Multilevel Approach to Accelerator Design for Reinforcement Learning on FPGA-Based Systems
  • Cite Icon11
  • https://doi.org/10.1109/tcad.2020.3028350Copy DOI Icon

ReLAccS: A Multilevel Approach to Accelerator Design for Reinforcement Learning on FPGA-Based Systems

  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Reinforcement learning (RL), specifically Q-learning, with human-like learning abilities to learn from experience without any a priori data, is being increasingly used in embedded systems in the field of control and navigation. However, finding the optimal policy in this approach can be highly compute-intensive, and a software-only implementation may not satisfy the application's timing constraints. To this end, we propose optimization methods at multiple levels of accelerator design for RL. Specifically, at the architecture-level, we exploit the instruction-level parallelism and the spatial parallelism in FPGAs to improve the throughput over state-of-the-art designs by up to 34%. Further, we propose lookup table-level optimizations to reduce the resource utilization and power dissipation of the accelerator. Finally, we propose algorithm-level approximation that can be used for acceleration of Q-learning problems with more states and for reducing the peak power dissipation. We report up to 10× reduction in power dissipation with marginal degradation in quality of results.

Similar Papers
  • Research Article
  • Citations8

Different I/O Standard and Technology Based Thermal Aware Energy Efficient Vedic Multiplier Design for Green Wireless Communication on FPGA

  • Jun 01, 2017
  • Wireless Personal Communications
  • Kavita Goswami +3
  • Research Article
  • Citations4

A Quality-Configurable Approximate Serial Bus for Energy-Efficient Sensory Data Transfer

  • Sep 01, 2018
  • IEEE Journal on Emerging and Selected Topics in Circuits and Systems
  • Setareh Behroozi +3
  • Research Article
  • Citations2

Design of an energy efficient approximate BinDCT module in quantum cellular automata

  • Jun 05, 2025
  • Scientific Reports
  • Mohsen Vahabi +3
  • Conference Article

Power dissipation minimization in RF front ends

  • Sep 01, 2010
  • P G M Baltus +3
  • Research Article
  • Citations2

Low power design of variable centre frequency CMOS VCO

  • Dec 01, 2007
  • International Journal of Electronics
  • A Billore +2
  • Conference Article
  • Citations46

Reducing power dissipation after technology mapping by structural transformations

  • Jan 01, 1996
  • Bernhard Rohfleisch +2
  • Conference Article
  • Citations3

Reconfigurable low-power Concurrent Error Detection in logic circuits

  • Dec 01, 2010
  • Sobeeh Almukhaizim +2
  • Conference Article
  • Citations2

A Passive Reduced Rating Output Rectifier Snubber for Plasma Cutting Power Supply

  • Mar 19, 2006
  • G.R Kamath
  • Conference Article
  • Citations7

Power Reduction in FinFET Half Adder using SVL Technique in 32nm Technology

  • Jan 01, 2019
  • M N Satish +1
  • Conference Article
  • Citations5

Design and Analysis of Novel Interconnects with Network-on-Chip LVDS Transmitter for Low Delay

  • Aug 01, 2018
  • Jayshree +1
  • Conference Article
  • Citations7

Ultra wide band low noise amplifier with self-bias for improved gain and reduced power dissipation

  • Apr 01, 2016
  • Vivek Singh +1
  • Research Article
  • Citations16

Supply and power optimization in leakage-dominant technologies

  • Sep 01, 2005
  • IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
  • Man Lung Mui +2
  • Conference Article
  • Citations24

AXSERBUS: A quality-configurable approximate serial bus for energy-efficient sensing

  • Jul 01, 2017
  • Younghyun Kim +3
  • Conference Article
  • Citations22

Implementation of digital filters in carry-save residue number system

  • Jan 01, 2001
  • A Del Re +2
  • Research Article
  • Citations2

A 16-bit microprocessor on SOS–PULCE–

  • Jan 01, 1979
  • Japanese Journal of Applied Physics
  • Jun Iwamura +7
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.