Deep Learning (DL), a successful promising approach for discriminative and generative tasks, has recently proved its high potential in 2D medical imaging analysis; however, physiological data in the form of 1D signals have yet to be beneficially exploited from this novel approach to fulfil the desired medical tasks. Therefore, in this paper we survey the latest scientific research on deep learning in physiological signal data such as electromyogram (EMG), electrocardiogram (ECG), electroencephalogram (EEG), and electrooculogram (EOG). We found 147 papers published between January 2018 and October 2019 inclusive from various journals and publishers. The objective of this paper is to conduct a detailed study to comprehend, categorize, and compare the key parameters of the deep-learning approaches that have been used in physiological signal analysis for various medical applications. The key parameters of deep-learning approach that we review are the input data type, deep-learning task, deep-learning model, training architecture, and dataset sources. Those are the main key parameters that affect system performance. We taxonomize the research works using deep-learning method in physiological signal analysis based on: (1) physiological signal data perspective, such as data modality and medical application; and (2) deep-learning concept perspective such as training architecture and dataset sources.

Physics education applications using augmented reality technology, which has been developed extensively in recent years, have a lot of restrictions in terms of performance and accuracy. The purpose of our research is to develop a real-time simulation system for physics education that is based on parallel processing. In this paper, we present a video see-through AR (Augmented Reality) system that includes an environment recognizer using a depth image of Microsoft’s Kinect V2 and a real-time soft body simulator based on parallel processing using GPU (Graphic Processing Unit). Soft body simulation can provide more realistic simulation results than rigid body simulation, so it can be more effective in systems for physics education. We have designed and implemented a system that provides the physical deformation and movement of 3D volumetric objects, and uses them in education. To verify the usefulness of the proposed system, we conducted a questionnaire survey of 10 students majoring in physics education. As a result of the questionnaire survey, 93% of respondents answered that they would like to use it for education. We plan to use the stand-alone AR device including one or more cameras to improve the system in the future.

This research develops an effective and precise collision detection (CD) algorithm for real-time simulation in virtual environments such as computer graphics, realistic and immersive virtual reality (VR), augmented reality (AR) and physical-based simulation within an enhanced algorithm for object collision detection in 3D geometry. We describe an improved algorithm through a comparison in the application of a central processing unit (CPU) and graphics processing units (GPU). Although leveraging CPU for computational speed improvements has gained significant recognition in recent years, this study distinguishes by tracking 3D geometry bounding volume hierarchy (BVH) constructed in a spatial decomposition structure with a focus on Octree-based Axis-Aligned Bounding Box (AABB) structure in 3D scene to compute collision detection to swiftly reject disjoint objects and minimize the number of triangle primitives that need to be processed and then the Möller method is utilized to compute precise triangle primitives, further enhancing the efficiency and precision of the collision detection process. This approach is also designed to implement computation with GPU which utilizes the high-level shader language (HLSL) programming language on the compute shader Unity3D. AABB is structured as the maximum and minimum hexahedron enclosing an object that is parallel to the coordinate axis. Otherwise, GPU computational technique is a crucial method for further enhancing the object's performance. The proposed method utilizes Octree AABB-based GPU parallel processing to reduce the computational load of real-time collision detection simulations and to handle multiple computations simultaneously. Comparative performance evaluations demonstrate that our GPU-accelerated framework consistently reaches the fastest collision detection times from 1.01 to 45.62 times, respectively.

Real-time physics simulations are vital for creating immersive and interactive experiences in extended reality (XR) applications. Balancing computational efficiency and simulation accuracy is challenging, especially in environments with multiple deformable objects that require complex interactions. In this study, we introduce a GPU-based parallel processing framework combined with a position-based dynamics (PBD) solver to tackle these challenges. The system is deployed within the Unity engine and enhances real-time performance through the use of sophisticated collision detection and response algorithms. Our method employs an AABB-based bounding volume hierarchy (BVH) structure to efficiently detect collisions, and incorporates the Möller–Trumbore algorithm for precise triangle-level interactions. We also boost computational efficiency by storing collision data in GPU-accessible 2D textures. Experimental assessments show performance improvements of up to 1705% in GPU simulations over CPU counterparts, achieving stable real-time frame rates for complex models such as the Stanford Bunny and Armadillo. Furthermore, utilizing 2D texture storage improves the FPS by up to 117%, confirming its efficacy for XR applications. This study offers a robust, scalable framework for real-time physics simulations, facilitating more natural and immersive XR experiences.

This paper presents a comparative study of the Vertex Block Descent (VBD) and Position-Based Dynamics (PBD) algorithms, focusing on their performance in physical simulation tasks. Unity, a versatile physics engine, served as the simulation platform for the experiments. Among various types of physical simulations of deformable objects, fluids, and cloth dynamics, cloth simulations were chosen for implementation with both algorithms. The experimental setup ensured identical parameters, including time steps and movement behavior, for both algorithms across scenarios involving hanging, object-to-object collisions, and self-collisions. The results indicate that while the performance difference in frames per second (fps) between the two algorithms is negligible for simulations with a small number of nodes, the VBD algorithm consistently outperforms the PBD algorithm as the node count increases. Furthermore, this study provides practical guidelines for maintaining real-time performance, detailing the maximum node count each algorithm can support, while sustaining a minimum threshold of 30 fps, which is necessary for real-time applications. The comparison was conducted using CPU-based computation to establish a baseline for future studies in GPU-accelerated environments, where parallel processing is expected to further highlight the performance advantages of VBD. Future work will extend this research by evaluating additional physical simulation models, including the Mass-Spring System and Extended Position-Based Dynamics (XPBD), and developing optimizations to enhance the efficiency and scalability of these algorithms.