IEEE Transactions on Human-Machine Systems最新文献

{"title":"Multidomain EEG Representation Learning With Orthogonal Mapping and Attention-Based Fusion for Cognitive Load Classification","authors":"Prithila Angkan;Amin Jalali;Paul Hungler;Ali Etemad","doi":"10.1109/THMS.2026.3651706","DOIUrl":"https://doi.org/10.1109/THMS.2026.3651706","url":null,"abstract":"We propose a new representation learning solution for the classification of cognitive load based on electroencephalogram (EEG). Our method integrates both time and frequency domains by first passing the raw EEG signals through the convolutional encoder to obtain the time domain representations. Next, we measure the power spectral density for all five EEG frequency bands and generate the channel power values as 2-D images referred to as multispectral topography maps. These multispectral topography maps are then fed to a separate encoder to obtain the representations in frequency domain. Our solution employs a multidomain attention module that maps these domain-specific embeddings onto a shared embedding space to emphasize more on important interdomain relationships to enhance the representations for cognitive load classification. In addition, we incorporate an orthogonal projection constraint during the training of our method to effectively increase the interclass distances while improving intraclass clustering. This enhancement allows efficient discrimination between different cognitive states and aids in better grouping of similar states within the feature space. We validate the effectiveness of our model through extensive experiments on two public EEG datasets, CL-Drive and CLARE for cognitive load classification. Our results demonstrate the superiority of our multidomain approach over the traditional single-domain techniques. Moreover, we conduct ablation and sensitivity analyses to assess the impact of various components of our method. Finally, robustness experiments on different amounts of added noise demonstrate the stability of our method compared to other state-of-the-art solutions.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"343-355"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Guided Self-Attention: Find the Generalized Necessarily Distinct Vectors for Grain Size Grading","authors":"Fang Gao;Xuetao Li;Jiabao Wang;Shengheng Ma;Jun Yu","doi":"10.1109/THMS.2026.3653470","DOIUrl":"https://doi.org/10.1109/THMS.2026.3653470","url":null,"abstract":"With the development of steel materials, metallographic analysis has become increasingly important. Unfortunately, grain size analysis is a manual process that requires experts to evaluate metallographic photographs, which is unreliable and time-consuming. To resolve this problem, we propose a novel classification method based on deep learning, namely, GSNets, a family of hybrid models, which can effectively introduce guided self-attention for classifying grain size. Concretely, we build our models from three insights: 1) Introducing our novel guided self-attention module can assist the model in finding the generalized necessarily distinct vectors capable of retaining intricate relational connections and rich local feature information; 2) By improving the pixelwise linear independence of the feature map, the highly condensed semantic representation will be captured by the model; 3) Our novel triple-stream merging module can significantly improve the generalization capability and efficiency of the model. Experiments show that our GSNet yields a classification accuracy of 90.1%, surpassing the state-of-the-art Swin Transformer V2 by 1.9% on the steel grain size dataset, which comprises 3599 images with 14 grain size levels. Furthermore, we intuitively believe our approach is applicable to broader applications such as object detection and semantic segmentation.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"224-236"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"IEEE Systems, Man, and Cybernetics Society Information","authors":"","doi":"10.1109/THMS.2026.3679068","DOIUrl":"https://doi.org/10.1109/THMS.2026.3679068","url":null,"abstract":"","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"C2-C2"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11475810","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148505127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Trust Process Applied to Machine Learning Algorithms: The Influence of Calibrated Confidence Estimates","authors":"Gene M. Alarcon;Sarah A. Jessup;Scott K. Meyers;Sasha M. Willis;Krista N. Harris;Dexter Johnson;Jacob Noblick;Walter D. Bennette","doi":"10.1109/THMS.2026.3651060","DOIUrl":"https://doi.org/10.1109/THMS.2026.3651060","url":null,"abstract":"Little research has been conducted on user interactions with machine learning models and users' trust. The current study explores differences in psychological perceptions of trust between two machine learning models. Across three studies, participants (<italic>N</i> = 130, 134, 132) were recruited online and completed an image-binning monitoring task with either an uncalibrated classification model (UCM) or a calibrated classification model (CCM) that provided confidence estimates about their decisions. The UCM was highly confident, regardless of images from out-of-distribution classes, whereas the CCM was better calibrated between accuracy and confidence of out-of-distribution images. Results revealed participants performed better on the task in the CCM condition across all studies. In addition, trustworthiness perception results differed depending on whether the study included out-of-distribution images. Overall, calibrated confidence estimates displayed by CCMs can be an effective means of increasing transparency and enhancing our understanding of trust in machines, but more research is needed on trust perceptions in image classification tasks.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"301-310"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148505246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Next Generation of Brain-Computer Interfaces: A Proposal for Spikes Detection Based on CDMA and Power Domain Nonorthogonal Multiple Access Techniques","authors":"Jesus Minguillon;Miguel Angel Lopez-Gordo;Pablo Martinez-Cañada","doi":"10.1109/THMS.2026.3663792","DOIUrl":"https://doi.org/10.1109/THMS.2026.3663792","url":null,"abstract":"The first generation of invasive brain-computer interfaces was conceived as a centralized and monolithic technology composed of an array of microelectrodes implanted on the brain surface and wired to an external unit for telemetry and power supply. Although this technology exceeded the expectations and possibilities of the noninvasive one, risks associated with implantation of wired devices and procedures appeared. Advances in electronics miniaturization and wireless power transfer methods, among others, have favored the proposal of wireless submillimeter implantable devices with stimulation and telemetry capabilities. This constitutes the foundation stone of a next generation of submilliliter implanted neurosensors (NG-SINs) based on massive distributed networks of untethered neuroimplants deployed on the brain. This particular ecosystem, because of the inherent severe restrictions in size, power consumption, electromagnetic safety, and computational capacity, presents significant challenges in the field of communications and internetworking, such as media access control, addressing, data error rate, and others. In this study, we propose a combination of orthogonal multiple access (OMA) and nonorthogonal media access (NOMA) as the base for the telemetry of the NG-SINs, with a focus on the detection of neural spikes in real time. Specifically, we propose the use of asynchronous code division multiple access (aCDMA) and power domain non orthogonal media access (PD-NOMA) with successive interference cancelation (SIC) over backscattered communications. While aCDMA mimics the spontaneous nature of brain activity, and provides addressing and grant-free random-access mechanism, its inherent near-far issue can be conveniently used to improve the performance by means of PD-NOMA and SIC. Based on our simulations, we conclude that this combination of NOMA and OMA techniques is a viable and promising approach for the NG-SINs applied to neural spikes detection that enables scalability with improved performance.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"281-290"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11415676","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148505465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fully Transparent Shape-Changing Display With Adjustable Key Travel for Enhanced Touchscreen Typing","authors":"Pengbo Zhao;Yuhui Wang;Boxue Shan;Yan Liu;Yibin Wang;Yuru Zhang;Dangxiao Wang","doi":"10.1109/THMS.2026.3651819","DOIUrl":"https://doi.org/10.1109/THMS.2026.3651819","url":null,"abstract":"Touchscreen devices often suffer from high typing error rates in the absence of physical keys. Shape-changing displays, which recreate physical keys on the screen, offer a promising solution. While the travel of physical keys plays an important role in the typing experience, it has received limited attention in previous work. To investigate the impact of on-screen physical keys with varying key travel on typing performance and user preferences, we developed a fully transparent shape-changing display that delivers adjustable key travel on touchscreens. The engineered portable microfluidic pressure-control system regulates the driving pressure, thereby adjusting the key travel to provide personalized passive tactile feedback. Using this platform, we conducted a user study in which participants typed with a two-thumb posture under four different key travel conditions, including the zero-travel condition. The results demonstrate that shape-changing keys significantly reduce typing error rates compared with flat keys, while revealing significant inter-individual variability in both performance and preference. These findings highlight the role of adjustable key travel in meeting diverse user needs and enabling adaptive touchscreen interaction.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"390-399"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148506072","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Human-Autonomy Teaming in Robot Navigation Using Bioinspired Approach for Search and Rescue","authors":"Timothy Sellers;Tingjun Lei;Chaomin Luo;Erfu Yang;Zhuming Bi","doi":"10.1109/THMS.2026.3651248","DOIUrl":"https://doi.org/10.1109/THMS.2026.3651248","url":null,"abstract":"Search and rescue operations save thousands of lives yearly, and robotics can further improve survival rates in hazardous, partially observable environments. This article introduces a trustworthy human-autonomy teaming (HAT) navigation framework that fuses human situational awareness with algorithmic planning across both global and local layers. At the global level, a hybrid graph integrates generalized Voronoi diagrams (GVDs), for high-clearance, topology-preserving roadmaps interpretable to operators, with a spatiotemporal graph that dynamically updates edges and node salience as obstacles and mission priorities evolve. Human input is incorporated through a node optimization protocol that prioritizes access nodes, adjusts waypoint sequences, and invokes RRT*-informative path planning when jump-search shortcuts become unreliable, thereby aligning exploration with operator-defined areas of interest and mission constraints. At the local level, a bioinspired neural network (BNN) with an adaptive window strategy provides reactive obstacle avoidance and target tracking, ensuring smooth and safe motion in cluttered, dynamic scenes. Simulation results across indoor and urban-like environments demonstrate that the proposed framework enhances path efficiency, robustness to environmental change, and responsiveness while reducing unnecessary re-planning. By unifying GVD-based global structure, human-guided decision-making, and BNN-based local reactivity, the HAT framework advances autonomous navigation toward greater safety, adaptability, and effectiveness in real-world search and rescue operations.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"311-320"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504582","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Text Mining Analysis of 2809 Articles on Fatigue Data Acquisition Methods for Traffic-Related Personnel: A Systematic Review Leveraging Natural Language Models","authors":"Shengjian Hu;Weining Fang;Haifeng Bao","doi":"10.1109/THMS.2026.3651296","DOIUrl":"https://doi.org/10.1109/THMS.2026.3651296","url":null,"abstract":"Fatigue among transportation-related personnel is a critical factor impacting traffic safety. In fatigue detection studies, data acquisition plays an essential role in determining the accuracy and reliability of the results. The data acquisition methods used in existing studies vary significantly and include electroencephalography, eye activity monitoring, and heart rate change analysis, each with distinct advantages and limitations. Such methodological diversity necessitates a systematic evaluation of their applicability and performance. Given the growing body of research in fatigue detection, traditional qualitative reviews face considerable challenges in covering the breadth of available information. To address this gap, this article proposes a structured information extraction framework leveraging a large language model to systematically evaluate 2809 studies on fatigue detection published between 2000 and May 2025. The developed framework achieved high accuracy in article screening (95.62%) and demonstrated robust performance in generating structured text. Our findings show an almost equal distribution between contact-based (50.24%) and noncontact-based (49.76%) detection methods, with parallel developmental trends and a strong positive correlation (r = 0.9569). Temporal trends indicate substantial research growth after 2012, coinciding with advancements in sensor technologies and computational capabilities. Key methodological combinations include eye–face signals (186 instances) for behavioral detection and brain–heart signals (102 instances) for physiological monitoring. Our findings serve as a foundational reference for selecting appropriate methods, and the proposed framework demonstrates potential applicability across diverse scientific disciplines.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"267-280"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504585","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multimodal Feedback in Automated Training Systems for Cardiopulmonary Resuscitation: A Review","authors":"Mine Sarac;Sevval Cetin;Baran Aslan;Ali Tas;Fabio Stroppa","doi":"10.1109/THMS.2025.3641695","DOIUrl":"https://doi.org/10.1109/THMS.2025.3641695","url":null,"abstract":"Incorporating multimodal feedback in automated cardiopulmonary resuscitation (CPR) training systems has emerged as a crucial advancement in medical education, aiming to improve the quality and efficacy of CPR performance. This work presents an extensive and technical literature search on automated training CPR systems, specifically focusing on the feedback modality they provide during training sessions. We completed the search in the IEEE, ACM, Springer, SAGE, Elsevier, MDPI, Scholar, and Scopus databases between 2015 and 2025 (August) using the keywords “CPR,” “Virtual Reality (VR),” “Augmented Reality (AR),” “feedback,” and “training.” We categorized our findings based on the type of feedback provided (i.e., visual, audio, haptic, or a combination of these), the display type used to render the feedback, the technology used to monitor the trainee’s performance, and the type of manikin used. We conclude with recommendations for future research.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"410-424"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148506703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effects of NDRT Interface Display and Control Modalities on Motion Sickness and Preference During Autonomous Driving","authors":"Sungmin Kim;Kitae Hwang;Woojin Park","doi":"10.1109/THMS.2026.3653811","DOIUrl":"https://doi.org/10.1109/THMS.2026.3653811","url":null,"abstract":"This study investigates the impact of nondriving-related task (NDRT) interface design, specifically the choice of display and control modalities, on the severity of autonomous vehicle motion sickness (AV MS) and passenger preference. Participants completed a letter 2-back task under four different interface conditions: visual-manual, visual-speech, auditory-manual, and auditory-speech. The Motion Sickness Assessment Questionnaire was used to measure MS levels before and after each on-road trial. Results indicated that MS severity was significantly lower for auditory display than for visual display, while control modality had no significant effect. In terms of preference, participants showed a higher preference for visual interfaces in the in-laboratory environment, whereas auditory and speech control interfaces were preferred in the on-road environment. These findings suggest that optimizing NDRT interface design, particularly through the use of auditory displays and speech controls, can mitigate AV MS and enhance passenger comfort. This research contributes to the ongoing efforts to improve user experience and acceptance of AV technology.","PeriodicalId":48916,"journal":{"name":"IEEE Transactions on Human-Machine Systems","volume":"56 2","pages":"247-256"},"PeriodicalIF":4.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504874","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}