BIROn - Birkbeck Institutional Research Online

    Robustness and performance of Deep Reinforcement Learning

    Al-Nima, R. and Han, Tingting and Al-Sumaidaee, S. and Chen, Taolue and Woo, W. (2021) Robustness and performance of Deep Reinforcement Learning. Applied Soft Computing 105 (107295), ISSN 1568-4946.

    [img]
    Preview
    Text
    Robustness and Performance of Deep Reinforcement Learning.pdf - Author's Accepted Manuscript
    Available under License Creative Commons Attribution Non-commercial No Derivatives.

    Download (3MB) | Preview

    Abstract

    Deep Reinforcement Learning (DRL) has recently obtained considerable attention. It empowers Reinforcement Learning (RL) with Deep Learning (DL) techniques to address various difficult tasks. In this paper, a novel approach called the \emph{Genetic Algorithm of Neuron Coverage} (GANC) is proposed. It is motivated for improving the robustness and performance of a DRL network. The GANC uses Genetic Algorithm (GA) to maximise the Neuron Coverage (NC) of a DRL network by producing augmented inputs. We apply this method in the self-driving car applications, where it is crucial to accurately provide a correct decision for different road tracking views. We evaluate our method on the SYNTHIA-SEQS-05 databases in four different driving environments. Our outcomes are very promising - the best driving accuracy reached 97.75% - and are superior to the state-of-the-art results.

    Metadata

    Item Type: Article
    Keyword(s) / Subject(s): Deep Reinforcement Learning, Genetic Algorithm, Neuron Coverage, Road tracking
    School: School of Business, Economics & Informatics > Computer Science and Information Systems
    Depositing User: Tingting Han
    Date Deposited: 31 Mar 2021 15:00
    Last Modified: 16 Mar 2022 01:10
    URI: https://eprints.bbk.ac.uk/id/eprint/43509

    Statistics

    Activity Overview
    6 month trend
    32Downloads
    6 month trend
    91Hits

    Additional statistics are available via IRStats2.

    Archive Staff Only (login required)

    Edit/View Item Edit/View Item