The natural locomotion interface is critical to the development of many VR applications. For household applications, there are two basic requirements: immersive experience and minimized space occupation. existing strategies generally do not simultaneously satisfy these requirements well. This article presents a novel omnidirectional treadmill (ODT) system named Hex-Core-MK1 (HCMK1). By implementing kinds mirror-symmetrical spiral rollers generate velocity field, this proposed capable...

Learning about the cognitive state of brain has always been a popular topic. Based on fact that fluctuations signals and functional connectome (FC) relate to specific human behaviors, deep learning based methods have shown promising results prediction such behaviors by analyzing biological signals. Existing either model from static perspectives or apply spatial-temporal graph convolution extract dynamic properties. However, information can reflect global activities local respectively. Thus,...

PurposeTo validate the effectiveness of an approach called batch-balanced focal loss (BBFL) in enhancing convolutional neural network (CNN) classification performance on imbalanced datasets.Materials and MethodsBBFL combines two strategies to tackle class imbalance: (1) batch-balancing equalize model learning samples (2) add hard-sample importance gradient. BBFL was validated fundus image datasets: a binary retinal nerve fiber layer defect (RNFLD) dataset (n = 7,258) multiclass glaucoma...

Abstract Efficient cascaded proton acceleration driven by an intense Laguerre–Gaussian (LG) laser is realized in combined three-dimensional particle-in-cell simulations and CST STUDIO SUITE (CST) simulations. show that there no divergent force component the transverse direction coil center. Therefore, collimated beam LG first stage benefits from uniform second stage. By contrast, with larger divergence disperses to outside of because near wire Gaussian case. Finally, a quasi-monoenergetic...

Abstract Purpose To clarify the causal relationship between factors contributing to postoperative survival of patients with esophageal cancer. Methods A cohort 195 who underwent surgery for cancer 2008 and 2021 was used in study. All had preoperative chest computed tomography (CT) positron emission tomography‐CT (PET‐CT) scans prior receiving any treatment. From these images, high throughput quantitative radiomic features, tumor various body composition features were automatically extracted....

Abstract Background Chest x‐ray is widely utilized for the evaluation of pulmonary conditions due to its technical simplicity, cost‐effectiveness, and portability. However, as a two‐dimensional (2‐D) imaging modality, chest images depict limited anatomical details are challenging interpret. Purpose To validate feasibility reconstructing three‐dimensional (3‐D) lungs from single 2‐D image via Vision Transformer (ViT). Methods We created cohort 2525 paired (scout images) computed tomography...

This study aims to establish the causal relationship network between various factors leading workday loss in underground coal mines using a novel artificial intelligence (AI) method. The analysis utilizes data obtained from National Institute for Occupational Safety and Health (NIOSH). A total of 101,010 injury records 3,982 unique spanning years 1990 2020 were extracted NIOSH database. Causal relationships analyzed visualized AI method called Grouped Greedy Equivalence Search (GGES). impact...

Abstract Background: Pulmonary function tests (PFTs) and computed tomography (CT) imaging are vital in diagnosing, managing, monitoring lung diseases. A common issue practice is the lack of access to recorded pulmonary functions despite available chest CT scans. Purpose: To develop validate a deep learning algorithm for predicting directly from Methods: The development cohort came Pittsburgh Lung Screening Study (PLuSS) (n=3619). validation Specialized Centers Clinically Oriented Research...

Multi-modal pre-trained models efficiently extract and fuse features from different modalities with low memory requirements for fine-tuning. Despite this efficiency, their application in disease diagnosis is under-explored. A significant challenge the frequent occurrence of missing modalities, which impairs performance. Additionally, fine-tuning entire model demands substantial computational resources. To address these issues, we introduce Modality-aware Low-Rank Adaptation (MoRA), a...

Large language models demonstrate impressive performance on downstream tasks, yet requiring extensive resource consumption when fully fine-tuning all parameters. To mitigate this, Parameter Efficient Fine-Tuning (PEFT) strategies, such as LoRA, have been developed. In this paper, we delve into the concept of task-specific directions--critical for transitioning large from pre-trained states to enhancements in PEFT. We propose a framework clearly define these directions and explore their...

In clinical In-Vitro Fertilization (IVF), identifying the most viable embryo for transfer is important to increasing likelihood of a successful pregnancy. Traditionally, this process involves embryologists manually assessing embryos' static morphological features at specific intervals using light microscopy. This manual evaluation not only time-intensive and costly, due need expert analysis, but also inherently subjective, leading variability in selection process. To address these...

With the rapid development of remote sensing technology and computer technology, machine learning algorithms are widely used in optical bathymetry inversion. However, due to many parameters algorithm, parameter selection has a large impact on accuracy model To address this problem, paper adopts Bayesian optimization algorithm select clerk forest model. The experimental results show that hyperparameter-based random highest accuracy, average relative error is only 10.60% which better than...

This paper presents a novel Linear Mapping Field (LMF) to map time series into two-dimensional images. The LMF extracts deeper features of fNIRS signals, which makes less reliant on some prior. developed convolution neural networks detect more prominent than the state-of-the-art methods. experimental results indicate that different from RNNs can only perceive in "sequential" manner, LMF's characteristic "jumpingly" perception is key achieving excellent results. <italic...

The application of reinforcement learning (RL) in artificial intelligence has become increasingly widespread. However, its drawbacks are also apparent, as it requires a large number samples for support, making the enhancement sample efficiency research focus. To address this issue, we propose novel N-step method. This method extends horizon agent, enabling to acquire more long-term effective information, thus resolving issue data inefficiency RL. Additionally, can reduce estimation variance...

Radio Frequency Fingerprinting (RFF) technology has emerged as a powerful solution for access control in wireless networks. One of the main tasks this process is removing channel interference to obtain reliable RFF features. This paper proposes novel extraction scheme based on reciprocity. When uplink and downlink channels satisfy short-term reciprocity, method calculates features by estimating Channel State Information (CSI) dividing them. It can eliminate multipath characteristics greatest...