Salvato in:
Dettagli Bibliografici
Autori principali: Ivanov, Dmitry A., Larionov, Denis A., Maslennikov, Oleg V., Voevodin, Vladimir V.
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2405.07748
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917664324059136
author Ivanov, Dmitry A.
Larionov, Denis A.
Maslennikov, Oleg V.
Voevodin, Vladimir V.
author_facet Ivanov, Dmitry A.
Larionov, Denis A.
Maslennikov, Oleg V.
Voevodin, Vladimir V.
contents In real applications of Reinforcement Learning (RL), such as robotics, low latency and energy efficient inference is very desired. The use of sparsity and pruning for optimizing Neural Network inference, and particularly to improve energy and latency efficiency, is a standard technique. In this work, we perform a systematic investigation of applying these optimization techniques for different RL algorithms in different RL environments, yielding up to a 400-fold reduction in the size of neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2405_07748
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Neural Network Compression for Reinforcement Learning Tasks
Ivanov, Dmitry A.
Larionov, Denis A.
Maslennikov, Oleg V.
Voevodin, Vladimir V.
Machine Learning
In real applications of Reinforcement Learning (RL), such as robotics, low latency and energy efficient inference is very desired. The use of sparsity and pruning for optimizing Neural Network inference, and particularly to improve energy and latency efficiency, is a standard technique. In this work, we perform a systematic investigation of applying these optimization techniques for different RL algorithms in different RL environments, yielding up to a 400-fold reduction in the size of neural networks.
title Neural Network Compression for Reinforcement Learning Tasks
topic Machine Learning
url https://arxiv.org/abs/2405.07748