This article presents Physics-Informed Neural Networks (PINNs), which integrate physical laws into neural network training to model complex systems governed by partial differential equations (PDEs). PINNs enhance data efficiency, allowing for accurate predictions with less training data, and have applications in fields such as biomedical engineering, geophysics, and material science. Despite their advantages, PINNs face challenges like learning high-frequency components and computational overhead. Proposed solutions include causality constraints and improved boundary condition handling. A numerical experiment demonstrates the effectiveness of PINNs in solving the one-dimensional heat conduction equation, showcasing enhanced model stability and accuracy. Overall, PINNs represent a significant advancement in merging machine learning with physics.

Show more

With the rapid development of artificial intelligence, non-player characters (NPCs) in games have become increasingly intelligent. NPCs undoubtedly enhance players’ gaming experience by allowing for better adjustments to game difficulty and narrative development through the optimization of NPC behavior logic. However, research on intelligent NPCs in games remains underdeveloped, highlighting the need for more systematic studies. To address this issue, we review the literature on intelligent NPCs and summarize the progress and trends of intelligent NPCs in games. We compare classical game NPCs with intelligent NPCs and conclude that continued development is essential for advancing intelligent NPCs. We should refine the scoring standards for NPC behaviors based on the content of different games, encouraging AI to exhibit human-like behaviors and enhance authenticity. Moreover, we should create an ideal environment for AI training to make the process more efficient and the results more effective. Furthermore, we propose the goals and challenges in the current development of NPCs. Finally, we summarize the progress and trends of intelligent NPCs in games to provide insights for future researchers.

Show more

This study targets multimodal data preprocessing and community detection in social media analysis. The research problem is to extract valuable insights from multimodal data and decipher social network structures for efficient information spread. The research employs a literature study, case examples, and algorithmic analysis to investigate the preprocessing steps for text and images and the Louvain algorithm for community detection. Results show proper preprocessing boosts data fusion, while community detection reveals network patterns, aiding marketing and public opinion management. Despite challenges, these techniques enhance data utility through solutions like multi-resolution algorithms, shaping better social media services and a more stable network environment.

Show more

The monitoring of oxygen saturation (SpO2) for premature babies carries great importance, due to the highly required accuracy of oxygen therapy, which is used to prevent premature babies’ dyspnea. Motion artifacts, however, can influence the quality of the photo-plethysmograph (PPG) signal, causing imprecision. Through this study, the accuracy of PPG detection for newborns can be improved by applying the Kalman filter to restitute motion artifacts by using an accelerator and a gyroscope to build reference signals and adaptively adjust the parameters of the Kalman filter. A quantitative study is used in this investigation with relative simulations of Heartbeat Rate (HR) estimation based on filtered PPG signals. The results, on average, show an absolute error of 1.69 BPM, a relative error of 1.82%, and a standard deviation of 9.37 BPM. The great accuracy of the algorithm can improve SpO2 monitoring for newborns, further lowering the working pressure for medical workers. On the other hand, the standard deviation is relatively high, indicating that further work, such as improving the preprocess of signals or introducing machine learning, should be paid attention to.

Show more