KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Settings

Start date: 2024Subject: search.filters.subject.Deep reinforcement learning

Search Results

Now showing 1 - 2 of 2
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    A Comparative Study of Deep Reinforcement Learning Agents for Gold Trading with Technical Indicators and LLM-Filtered News Sentiment
    (2026-06-16)
    Thanasarn, Thanapong
    ;
    Anuntachai, Anuntapat
    ;
    Netisopakul, Ponrudee
    Integrating macroeconomic news into deep reinforcement learning (DRL) for daily gold (XAU/USD) trading remains challenging. This study implements a natural language processing pipeline using majority voting from three large language models (Llama-3.1-8B-Instruct, Qwen3-8B, and Gemma-3-12B-it) to filter articles from The New York Times from 2010 to 2024, yielding 592 relevant articles that are subsequently assigned sentiment scores using FinBERT. We evaluate A2C, PPO, and SAC agents in a FinRL and Stable-Baselines3 framework using a four-fold walk-forward expanding-window protocol and Optuna hyperparameter tuning. The models are compared under two feature settings: (1) six technical indicators only and (2) technical indicators combined with sentiment features. Results show that incorporating LLM-filtered sentiment can modestly improve trading performance for some agents. PPO with sentiment achieves the best average cumulative return (10.05%) and Sharpe ratio (0.74), compared with its indicator-only version (9.80%, 0.72), and slightly outperforms the Buy-and-Hold (B&H) baseline (9.58%, 0.71). A2C also improves with sentiment (9.50% to 9.90%), while SAC shows no improvement in this setting. These findings suggest that LLM-filtered sentiment provides a modest benefit for some DRL agents in daily gold trading in our experiments.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    An overview of reinforcement learning-based approaches for smart home energy management systems with energy storages
    (2024-09-01)
    Pinthurat, Watcharakorn
    ;
    Surinkaew, Tossaporn
    ;
    Hredzak, Branislav
    The paper's state-of-the-art review focuses on an in-depth evaluation of smart home energy management systems which employ reinforcement learning-based methods to integrate energy storages. In order to optimize energy consumption and improve overall sustainability while maintaining technical and economic constraints, the paper first investigates the multi-faceted aspects of integrating energy storages into smart homes. Second, an overview of a smart home system and a theoretical background of reinforcement learning-based algorithms are given and discussed. Consequently, this study delves into the challenges and benefits of integrating energy storage, specifically looking at ways to lessen the impact of renewable sources’ intermittency, improve grid stability, and streamline efficient energy storage management. Thirdly, the paper highlights the beneficial features of smart home energy storage integration, including reduced costs, increased system resilience, and improved energy efficiency. Therefore, cutting-edge reinforcement learning-based methods utilized in smart home energy management systems that incorporate energy storage are thoroughly examined by evaluating their effectiveness and adaptability, taking into account both multi-agent and single-agent reinforcement learning-based methods. Finally, the study identifies potential research directions, including the development of hybrid reinforcement learning algorithms, integration of demand-side management strategies, and addressing privacy and security concerns in reinforcement learning-based smart home energy management systems. While some research has made use of single-agent reinforcement learning, smart home energy storage systems that use energy storages seldom use multi-agent reinforcement learning techniques. Researchers, practitioners, and policymakers will be able to use this work as a foundation to build smart, sustainable home energy systems.