High-dimensional always-changing environments constitute a hard challenge for current reinforcement learning techniques. Artificial agents, nowadays, are often trained off-line in very static and controlled conditions in simulation such that training observations can be thought as sampled i.i.d. from the entire observations space. However, in real world settings, the environment is often non-stationary and subject to unpredictable, frequent changes. In this paper we propose and openly release CRLMaze, a new benchmark for learning continually through reinforcement in a complex 3D non-stationary task based on ViZDoom and subject to several environmental changes. Then, we introduce an end-to-end model-free continual reinforcement learning strategy showing competitive results with respect to four different baselines and not requiring any access to additional supervised signals, previously encountered environmental conditions or observations


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Continual Reinforcement Learning in 3D Non-stationary Environments



    Erscheinungsdatum :

    2020-01-01



    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629



    Hıerarchıcal reınforcement learnıng on non-statıonary envıronments

    Yücesoy, Yiğit Efe | BASE | 2015

    Freier Zugriff

    Continual SLAM: Beyond Lifelong Simultaneous Localization and Mapping Through Continual Learning

    Vödisch, Niclas / Cattaneo, Daniele / Burgard, Wolfram et al. | Springer Verlag | 2023


    CONTINUAL PROACTIVE LEARNING FOR AUTONOMOUS ROBOT AGENTS

    DAVIDSON JAMES / MASON JULIAN / POURSOHI ARSHAN | Europäisches Patentamt | 2021

    Freier Zugriff

    CONTINUAL PROACTIVE LEARNING FOR AUTONOMOUS ROBOT AGENTS

    DAVIDSON JAMES / MASON JULIAN / POURSOHI ARSHAN | Europäisches Patentamt | 2022

    Freier Zugriff

    Evaluating Differential Privacy in Federated Continual Learning

    Ouyang, Junyan / Han, Rui / Liu, Chi Harold | IEEE | 2023