IEEE Transactions on Emerging Topics in Computing最新文献

{"title":"MC-DiffNet: Mask-Constrained and Cycle-Consistent Diffusion Network for Unpaired Ultrasound Image Enhancement","authors":"Nuo Chen;Yongquan Zhang;Guolin Chen;Jinhao Zhao;Changmiao Wang;Amirmehdi Yazdani;Hai Wang","doi":"10.1109/TETC.2026.3695050","DOIUrl":"https://doi.org/10.1109/TETC.2026.3695050","url":null,"abstract":"Ultrasound imaging is a valuable tool in clinical diagnostics due to its real-time feedback and non-invasive nature. Despite these advantages, it often suffers from issues like low contrast, speckle noise, and blurred anatomical boundaries, which can compromise the accuracy of diagnoses. Although diffusion models have proven effective for image restoration, their use in enhancing unpaired ultrasound images is limited. Another limitation is lacking of structural constraints and region-specific guidance. To address these challenges, we introduce <bold>MC-DiffNet</b>, a multi-task, diffusion-based enhancement framework designed specifically for clinical ultrasound images in unpaired settings. Furthermore, the use of lesion masks provides structural guidance that enables the model to focus on pathological characteristics in ultrasound images. The framework incorporates three principal modules: a Cycle-Consistency Path, a Mask-Constrained Module, and a Context-Aware Classification Path. The Cycle-Consistency Path ensures consistency in unpaired training through degradation-reconstruction consistency. The Mask-Constrained Module incorporates lesion-aware structural constraints into the Structural Similarity Index Measure (SSIM) loss, aiming to maintain anatomical accuracy. Meanwhile, the Context-Aware Classification Path guides the enhancement process using semantic-level features. Together, these modules enable MC-DiffNet to enhance diagnostically significant regions with precision while preserving anatomical integrity. Experiments conducted on public ultrasound datasets reveal that our method outperforms existing techniques in both peak signal-to-noise ratio and SSIM. Remarkably, when applied to real-world unpaired datasets, MC-DiffNet offers improved structural fidelity and visual clarity. Further evaluation suggests that the enhanced images generated by our framework are beneficial for subsequent segmentation tasks, underscoring the clinical relevance and robustness of our approach.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"678-690"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148507243","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"Deep Belief Network-Based Activity of Daily Living Monitoring for Fall Risk Prediction in Elderly","authors":"Deepika Mohan;Peter Han Joo Chong;Jairo Gutierrez;Mirza Mansoor Baig;Hui Li","doi":"10.1109/TETC.2026.3686005","DOIUrl":"https://doi.org/10.1109/TETC.2026.3686005","url":null,"abstract":"Despite advancements in healthcare and emerging technologies, falls among older adults remain a significant health issue. Recent research has increasingly focused on developing advanced monitoring methods, predicting, and preventing falls in this population. Achieving high performance in fall prediction requires a clear understanding of relevant features such as gait patterns, balance metrics, muscle strength, and environmental factors. Identifying these key indicators and incorporating data from wearable sensors, medical histories, and demographic information can significantly enhance predictive accuracy. This study proposes an intelligent fall prediction model that anticipates future falls in older adults by continuously monitoring their Activities of Daily Living (ADLs) and detecting abnormalities. The model uses a Deep Belief Network (DBN) that incorporates contrastive divergence for pre-training, backpropagation for fine-tuning, and the Adam Optimizer to minimize loss. Evaluation of the proposed model shows it achieved an accuracy of 91.67%, specificity of 100%, and sensitivity of 90.00% when compared to the Ground Truth (GT) and existing fall prediction approaches. These results suggest that advanced deep learning techniques can effectively assist in early fall risk prediction, potentially reducing the likelihood and severity of falls among older adults.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"543-556"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148507250","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"YAXY: A New Hybrid FPGA-Based Hardware Acceleration of TFHE","authors":"Yangfu Xu;Aijiao Cui;Xiangyu Guo;Yier Jin","doi":"10.1109/TETC.2026.3691777","DOIUrl":"https://doi.org/10.1109/TETC.2026.3691777","url":null,"abstract":"Fully Homomorphic Encryption (FHE) enables direct computation on ciphertexts, thereby ensuring data privacy during processing. However, its widespread application is impeded by high computational costs, with bootstrapping being a critical efficiency bottleneck. Torus FHE (TFHE) facilitates efficient evaluation of arbitrary Boolean functions via fast gate bootstrapping. This paper proposes YAXY, an FPGA-accelerated design for TFHE bootstrapping. It combines high-level synthesis (HLS) and register-transfer level (RTL) optimizations to improve performance. Our approach employs parameterized key unrolling and pipelined key transmission at the HLS level, coupled with a hierarchically pipelined RTL architecture designed to maximize parallelism in external product computations. Implemented on the Xilinx ZCU102 platform at 300 MHz, the design achieves a latency of 0.44 ms and a throughput of 2,273 bootstrappings per second. It offers the lowest-latency standalone FPGA TFHE bootstrapping. Compared to processor-based implementations, our accelerator exhibits superior speed-resource tradeoff. The results validate that HLS-RTL co-design can significantly enhance the practicality of FHE for privacy-preserving applications.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"649-661"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148507563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"LOC-SAC: A Lyapunov-Guided Online Collaborative Offloading Strategy With the SAC Algorithm","authors":"Peiyan Yuan;Ming Li;Xiaoyan Zhao;Chenyang Wang;Hu Jin","doi":"10.1109/TETC.2026.3687572","DOIUrl":"https://doi.org/10.1109/TETC.2026.3687572","url":null,"abstract":"Lyapunov optimization plays a pivotal role in stabilizing systems and reducing energy consumption for online collaborative computing offloading. However, existing studies have encountered the issue of instantaneous greediness during the transformation process based on Lyapunov optimization, primarily due to the continuous operations involved in queue length calculations and local optimizations. This study investigates the application of soft actor-critic in optimizing the objective function of Lyapunov-guided queue models, aiming to achieve long-term and globally optimal offloading decisions. Firstly, the Lyapunov optimization framework is employed to build the service queue model and derive the upper bound of the system objective function. Secondly, the system action of resource allocation and the state space are constructed based on the queue model. Moreover, a deep reinforcement learning problem involving queue change and a Lyapunov drift-plus-penalty is formulated to minimize the system energy consumption. Thirdly, a queue-based internal discount factor is integrated into the system reward function of the SAC algorithm, enabling the system to achieve long-term gains more efficiently. Finally, an adaptive cloud-edge service offloading strategy is proposed to ensure queue stability and minimize energy consumption without necessitating a locally optimal solution at each time slot. Experimental results demonstrate that the proposed strategy exhibits superior scalability and robustness compared to other state-of-the-art and baseline approaches in online collaborative service offloading systems.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"588-603"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148507959","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"Privacy-Preserving Outsourced Deep Neural Network Training: New Dual-Cloud Framework and Efficient Construction","authors":"Chen-Fan Chang;Kuan-Chun Huang;Bo-Jia Chen;Yu-Chi Chen","doi":"10.1109/TETC.2026.3676652","DOIUrl":"https://doi.org/10.1109/TETC.2026.3676652","url":null,"abstract":"Recently, deep neural networks (DNNs) have achieved excellent results in face recognition, natural language processing, and object classification. With the development of data science and the continuous increase in the size of datasets, the computing power and storage requirements for DNN training have also increased. Although GPUs and training methods have been continuously optimized for decades, the storage requirements remain a significant cost. To reduce the cost, datasets are often centrally placed in cloud storage for interactive training. Data security is undoubtedly the primary concern for such applications. Datasets for DNN training are usually collected by various institutions and often contain a lot of sensitive information. To prevent confidential information from leaking, data in cloud storage must be encrypted or protected. In this paper, we use secret sharing, a very lightweight method, to implement privacy-preserving outsourced DNN training. Using secret sharing, the data and models can be split into two shares and outsourced to two cloud service providers for training. Since the calculations are performed by the cloud service providers, the model trainer can be a device with limited computational power, such as a lightweight portable device.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"467-482"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148509354","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"An Efficient Ternary in-Memory Computing Architecture With Application in Phonocardiogram-Based Driver Stress Detection","authors":"Amirhossein Fathollahi;Nima Eslami;Mohammad Hossein Moaiyeri","doi":"10.1109/TETC.2026.3668394","DOIUrl":"https://doi.org/10.1109/TETC.2026.3668394","url":null,"abstract":"Traffic accidents remain a leading global cause of injury and fatalities, with driver stress identified as a significant contributing factor. Traditional stress detection methods, such as ECG-based systems, suffer from bulky designs and limited battery life, restricting their practicality for real-time wearable applications. To address these limitations, this paper presents a novel ternary in-memory computing (IMC) architecture for efficient PCG-based stress classification in power-constrained wearable devices. The proposed design introduces ternary full-adder and multiplier circuits based on emerging technologies, enabling energy-efficient neural network acceleration. Simulation results demonstrate that the proposed ternary full adder and multiplier designs reduce overall energy consumption by 98% on average compared to existing ternary counterparts. Furthermore, implementing a 1-D neural network for efficient stress classification, the proposed design achieves a 99% reduction in energy-delay-product (EDP). The results indicate that the proposed architecture is well-suited for power-constrained wearable devices in PCG-based stress detection applications.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"414-425"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148508544","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"Time Information-Enhanced Multi-Space Modeling Spatiotemporal Transformer Network for Traffic Flow Forecasting","authors":"Bowen Yang;Zhizhe Lin;Jinglin Zhou;Hai Xie;Youyi Song;Teng Zhou","doi":"10.1109/TETC.2026.3680507","DOIUrl":"https://doi.org/10.1109/TETC.2026.3680507","url":null,"abstract":"Accurate traffic prediction is crucial for urban traffic planning and management. Recent advancements in spatiotemporal models have significantly improved the modeling of complex spatiotemporal correlations for traffic flow prediction. However, spatial correlations in traffic networks typically consist of both static and dynamic components. Many existing methods employ a single graph learning strategy to jointly model these components, which may limit their flexibility in comprehensively capturing the diverse spatial interactions among nodes. Additionally, the dynamic graph generation mechanisms based on input windows primarily rely on traffic flow observations within a limited time frame. This dependence on short-term observational data may hinder the effective capture of long-term spatiotemporal dependencies. To overcome these limitations, we propose a time information-enhanced multi-space modeling spatiotemporal transformer network (STMT) for traffic prediction. Specifically, we elaborate on three different spatiotemporal transformers for temporal modeling, static graph spatiotemporal modeling, and dynamic graph spatiotemporal dependency modeling to comprehensively model spatiotemporal relationships. Then, we design a dynamic graph generator that introduces a memory unit to provide historical traffic patterns during dynamic graph generation. Additionally, we propose time-node embeddings to adapt to different traffic patterns for the nodes at various times. Extensive experiments on six traffic datasets demonstrate that STMT achieves superior performance compared to 17 baseline methods.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"483-499"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148509356","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"A New Quantum Efficient Fault Tolerant Parity Preserving Clifford + T ALU","authors":"Shekoofeh Moghimi;Mohammad Reza Reshadinezhad;Antonio Rubio","doi":"10.1109/TETC.2026.3696481","DOIUrl":"https://doi.org/10.1109/TETC.2026.3696481","url":null,"abstract":"Reversible logic is characterized by its feature of being a prominent application in low-power conventional computers and the most compatible logic in quantum computers. Quantum circuits executes quantum algorithms and are important in a new generation of advanced computing. However, they are highly error-prone and also subject to the use of limited and sensitive physical resources like qubits and quantum gates. This paper introduces a comprehensive fault tolerant quantum-based reversible circuit which can calculate a high variety number of arithmetic and logic operations. It uses both: parity preserving method for detecting one or an odd number of bit errors, and Clifford + T implementation for protecting circuits against decoherence faults. First, we propose a new fault-tolerant (parity preserving and Clifford + T) adder/subtractor and derive a quantum reversible Arithmetic and Logic Unit (ALU). These modules reach the most optimized extent of circuit line, ancilla inputs, and garbage outputs according to the logical analysis done in this work. This article focuses on reducing T-cost metrics in conjunction with enhancing the number of operations and quantum costs in proposed circuits compared to other counterparts. We propose 4-bit fault tolerant ALU derived from our proposed single-bit module to show the proposed circuit can be extended to any larger size. The proposed circuit is implemented in the Qiskit simulation tool, and output results confirm the accuracy of our design functionality. The proposed full adder/ subtractor achieves an average of 64% and 43%, and the proposed ALU circuit achieves an average of 38% and 27% improvement in T-count and T-depth, respectively compared to their counterparts. The proposed reversible quantum-optimized fault tolerant adder/subtractor and ALU units can be applied as the basic part of DSP systems and quantum processor units in quantum computing devices.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"718-733"},"PeriodicalIF":5.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11543196","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148236384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"IEEE Transactions on Emerging Topics in Computing Publication Information","authors":"","doi":"10.1109/TETC.2026.3696909","DOIUrl":"https://doi.org/10.1109/TETC.2026.3696909","url":null,"abstract":"","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"C2-C2"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11561822","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148507963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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":"Nomoformer: A Transformer-Based Approach for Pedestrian Trajectory Prediction With Non-Linear Motion Representation","authors":"Yuhao Qing;Yueying Wang;Hai Wang;Huaicheng Yan","doi":"10.1109/TETC.2026.3673699","DOIUrl":"https://doi.org/10.1109/TETC.2026.3673699","url":null,"abstract":"Multi-agent trajectory prediction plays a crucial role in various domains, including autonomous driving, uncrewed systems, and robotics. However, existing approaches have not fully addressed the complex interplay between motion patterns and multi-agent interaction features, making accurate trajectory prediction in complex scenarios particularly challenging. To address these limitations, we present Nomoformer, a transformer-based architecture for multi-agent trajectory prediction. Nomoformer predicts agent trajectories through joint analysis of motion representations and spatio-temporal interactions. Our approach first models individual agent dynamics by incorporating physical constraints derived from velocity and angular features. We mapped select motion features to the frequency domain to capture inherent relationships between periodic patterns and agent topology. To model inter-agent interactions, we developed a spatio-temporal graph topology using heterogeneous graphs with directed edge features and attention masks. Within this topology, we introduced Edge Encoder and Node Encoder modules that effectively establish long-range dependencies across edge features and temporal dimensions. Extensive experiments across multiple benchmarks demonstrated that Nomoformer achieved state-of-the-art performance in predicting future motion trajectories.","PeriodicalId":13156,"journal":{"name":"IEEE Transactions on Emerging Topics in Computing","volume":"14 2","pages":"456-466"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148508545","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}