Faculty of Science and Technology

Permanent URI for this community

https://repository.dusit.ac.th/handle/123456789/461

Browse

Browsing Faculty of Science and Technology by Author "Pannee Suanpang"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Optimizing Electric Vehicle Charging Recommendation in Smart Cities: A Multi-Agent Reinforcement Learning Approach.
    (World World Electric Vehicle Journal, 2024-06-14) Pannee Suanpang
    As global awareness for preserving natural energy sustainability rises, electric vehicles (EVs) are increasingly becoming a preferred choice for transportation because of their ability to emit zero emissions, conserve energy, and reduce pollution, especially in smart cities with sustainable development. Nonetheless, the lack of adequate EV charging infrastructure remains a significant problem that has resulted in varying charging demands at different locations and times, particularly in developing countries. As a consequence, this inadequacy has posed a challenge for EV drivers, particularly those in smart cities, as they face difficulty in locating suitable charging stations. Nevertheless, the recent development of deep reinforcement learning is a promising technology that has the potential to improve the charging experience in several ways over the long term. This paper proposes a novel approach for recommending EV charging stations using multi-agent reinforcement learning (MARL) algorithms by comparing several popular algorithms, including the deep deterministic policy gradient, deep Q-network, multi-agent DDPG (MADDPG), Real, and Random, in optimizing the placement and allocation of the EV charging stations. The results demonstrated that MADDPG outperformed other algorithms in terms of the Mean Charge Waiting Time, CFT, and Total Saving Fee, thus indicating its superiority in addressing the EV charging station problem in a multi-agent setting. The collaborative and communicative nature of the MADDPG algorithm played a key role in achieving these results. Hence, this approach could provide a better user experience, increase the adoption of EVs, and be extended to other transportation-related problems. Overall, this study highlighted the potential of MARL as a powerful approach for solving complex optimization problems in transportation and beyond. This would also contribute to the development of more efficient and sustainable transportation systems in smart cities for sustainable development. Keywords: electric vehicle; charge station; multi-agent reinforcement learning; optimizing; recommendation systems; smart cities
  • Thumbnail Image
    Item
    Optimizing Tourism Service Intelligent Recommendation System By Multi-Agent Reinforcement Learning for Smart Cities Destination.
    (Operational Research in Engineering Sciences: Theory and Applications, 2023-11-14) Pannee Suanpang
    The tourism sector is in a state of continual evolution, marked by a growing demand from travellers for customized and individualized experiences within smart city destinations. In response to this evolving landscape, this research introduces an innovative approach to intelligent recommendation systems for tourism services, utilizing Multi-Agent Reinforcement Learning (MARL). The proposed methodology employs a centralized critic and decentralized actor architecture to capture intricate interactions among agents, thereby generating recommendations of superior quality. Performance evaluation conducted on a real-world dataset demonstrates the method's superiority over existing approaches in terms of recommendation accuracy and diversity. Furthermore, this paper introduces a tourism service recommendation system based on MARL and assesses its efficacy using five distinct algorithms: Real, Random, DQN, DDPG, and MADDPG. Results indicate that the MADDPG algorithm surpasses other algorithms in providing reliable, efficient, and cost-effective services to tourists. MADDPG's capacity to learn and adapt to shifting user preferences and behaviours, facilitated by a centralized critic and decentralized actors learning from agent-environment interactions, enables it to adeptly navigate complex and dynamic environments. Moreover, the research delves into the implications of these findings for the tourism industry, drawing insights from feedback obtained from 400 respondents. The results reveal a high degree of user satisfaction with the optimized tourism service recommendation system in smart city destinations, consequently fostering a strong intention among users to revisit. This study represents a notable advancement in augmenting the tourism experience through sophisticated recommendation systems tailored for smart city destinations.
  • Thumbnail Image
    Item
    SPSO-EFVM: A Particle Swarm OptimizationBased Ensemble Fusion Voting Model for Sentence-Level Sentiment Analysis
    (IEEE ACCESS, 2024-06-13) Pannee Suanpang
    T Sentiment analysis has received incremental growth in recent years for emerging applications, including human-robot integration, social platforms monitoring, and decision-support systems. Several neural or transformer model-based solutions have been provided in the field of sentiment analysis that relies on the decision of a single classifier or neural model. These are erroneous to encode contextual information into appropriate dialogues and increase extra computational cost and time. Hence, we proposed a compact and parameter-effective Particle Swarm Optimization-based Ensemble Fusion Voting Model (PSO-EFVM) that exploited the combined properties of four ensemble techniques, namely Adaptive-Boost, GradientBoost, Random-Forest, and Extremely-Randomized Tree with Particle Swarm Optimization (PSO)-based hyperparameter selection. The proposed model is investigated on five cross-domain datasets after applying the foremost initialization and feature extraction using Information Gain (IG). It employs adaptive and gradient learning to incorporate the automatic attribute selection with the arbitrary loss function optimization. In short, a generalized two-block composite classifier is designed to perform context compositionality and sentiment classification. A population-based meta-heuristic optimization PSO is applied to each base ensemble learner that calculates weights for the best parameter selection. Comprehensive investigations of different domains reveal the superiority of the proposed PSO-EFVM over established baselines and the latest state-of-the-art models.