Название: Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing 2022-Winter
Автор: Roger Lee
Издательство: Springer
Серия: Studies in Computational Intelligence
Год: 2023
Страниц: 165
Язык: английский
Формат: pdf (true), epub
Размер: 26.3 MB

This edited book presents scientific results of the 24th ACIS International Winter Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD2022-Summer) which was held on December 7–9, 2022, at Taichung, Taiwan. The aim of this conference was to bring together researchers and scientists, businessmen and entrepreneurs, teachers, engineers, computer users, and students to discuss the numerous fields of computer science and to share their experiences and exchange new ideas and information in a meaningful way. The conference organizers selected the best papers from those papers accepted for presentation at the workshop. The papers were chosen based on review scores submitted by members of the program committee and underwent further rigorous rounds of review. From this second round of review, 15 of the most promising papers are then published in this Springer (SCI) book and not the conference proceedings.

Reinforcement Learning (RL) is a powerful tool and has been increasingly used in continuous control tasks such as locomotion and balancing in robotics. In this paper, we tackle a balancing task in a highly dynamic environment, using a humanoid robot agent and a balancing board. This task requires complex continuous actuation in order for the agent to stay in a balanced state. In this work, we propose an RL algorithm structure based on the state-of-the-art Proximal Policy Optimization (PPO) using GPU-based implementation; the agent achieves successful balancing in under 40 min of real-time. We sought to examine the impact of action space shaping on sample efficiency and designed 6 distinct control modes. Our constrained parallel control modes outperform the naive baseline in both sample efficiency and variance to the starting seed.

As the supply and demand instability gradually increases due to the increase in obstacles to the power supply and the validity of predicting power demand, it is necessary to reduce or distribute power demand through management in terms of demand in parallel with the expansion of supply capacity. Accordingly, a paradigm shift in power policy was required due to efficient supply and demand including demand management. This study seeks to find a model suitable for the characteristics of power consumption per minute using Machine Learning and to identify optimal conditions through comparative experiments on matters to be considered when constructing a predictive model. LSTM prediction model was developed using Python code-based libraries. The ARIMA prediction model is developed using the IBM SPSS Statistics Version 22. Statistical Analysis Tool. For the evaluation of the predictive model, the Mean Absolute Percentage Error (MAPE), which is frequently used in the time series model, was used as the main performance indicator. This study aims to evaluate the results of the model by using the Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) as auxiliary indicators.

Contents: