Fundus diseases are prevalent causes of visual impairment and blindness worldwide, particularly in regions with limited access to ophthalmologists for timely diagnosis. Current approaches to fundus disease diagnosis heavily rely on expert-annotated data and AI-assisted image analysis, offering advantages, such as improved accuracy and accessibility. However, the dependency on annotated data poses a significant challenge, especially in regions with limited resources. To address this challenge, we propose a label-free general framework based on self-supervised machine learning. We performed feature distillation on a large number of unlabeled fundus images and employed a linear classifier for the detection of different fundus diseases. In validation experiments on the public and external validation fundus data sets, our model surpassed existing supervised approaches, achieving a remarkable increase in the area under the curve (AUC) of 15.7%, and even outperformed individual human experts. Our approach offers a promising solution to the limitations of current diagnostic methods, enhancing the potential for early and accurate detection of fundus diseases in resource-constrained settings.

Recently, the bit-serial-based processing-in-memory (PIM) has evolved as a promising solution to enhance the computing performance of data-intensive applications, due to its high performance and programmability. However, it is absent that a compiler can automatically convert an arbitrary Boolean function (generic workload) into PIM instructions, with optimized scheduling w.r.t. the varying hardware resource and specification. To fill the gap, we develop a logic compiler for bit-serial-based PIM (PIMLC). In PIMLC, we propose a workload-resource-aware scheduling to minimize the execution latency of a given parallel workload. Thanks to PIMLC, PIM can achieve$15.55\times$and$19.03\times$speedup (geo-mean) for SRAM- and ReRAM-PIM respectively, compared to the naive scheduling of prior work. PIMLC is publicly available at: https://github.com/Intelligent-Computing-Research-GroupIPIMLC.

Processing-in-Memory (PIM) has been widely explored for accelerating data-intensive machine learning computation that mainly consists of general-matrix-multiplication (GEMM), by mitigating the burden of data movements and exploiting the ultra-high memory parallelism. The two mainstreams of PIM, the analog- and digital-type, have both been exploited in accelerating machine learning workloads by numerous outstanding prior works. Currently, the digital-PIM is increasingly favored due to the broader computing support and the avoidance of errors caused by intrinsic non-idealities, e.g., process variation. Nevertheless, it still lacks further optimization considering the characteristics of the GEMM computation, including better efficient data layout and scheduling, and the ability to handle the sparsity of activations at the bit-level. To boost the performance and efficiency of digital SRAM PIM, we propose the architecture called VSPIM that performs the computation in a bit-serial fashion, with unique support of vector-scalar computing pattern. The novelties of the VSPIM can be concluded as follows: 1) support bit-serial based scalar-vector computing via ingenious parallel bit-broadcasting; 2) refine the GEMM mapping strategy and computing pattern to enhance performance and efficiency; 3) powered by the introduced scalar-vector operation, the bit-sparsity of activation is leveraged to halt unnecessary computation to maximize efficiency and throughput. Our comprehensive evaluation shows that, compared to the state-of-the-art SRAM-based digital-PIM design (Neural Cache), VSPIM can significantly boost the performance and energy efficiency by up to$8.87\times$and$4.81\times$respectively, with negligible area overhead, upon multiple representative neural networks.

Emotion recognition, a process of automatic cognition of human emotions, has great potential to improve the degree of social intelligence. Among various recognition methods, emotion recognition based on touch event’s temporal and force information receives global interests. Although previous studies have shown promise in the field of keystroke-based emotion recognition, they are limited by the need for long-term text input and the lack of high-precision force sensing technology, hindering their real-time performance and wider applicability. To address this issue, in this article, a piezoelectric-based keystroke dynamic technique is presented for quick emotion detection. The nature of piezoelectric materials enables high-resolution force detection. Meanwhile, the data collecting procedure is highly simplified because only the password entry is needed. International Affective Digitized Sounds (IADS) are applied to elicit users’ emotions, and a pleasure-arousal-dominance (PAD) emotion scale is used to evaluate and label the degree of emotion induction. A random forest (RF)-based algorithm is used in order to reduce the training dataset and improve algorithm portability. Finally, an average recognition accuracy of 79.33% of four emotions (happiness, sadness, fear, and disgust) is experimentally achieved. The proposed technique improves the reliability and practicability of emotion recognition in realistic social systems.

Sensor-based rehabilitation physical training assessment methods have attracted significant attention in refined evaluation scenarios. A refined rehabilitation evaluation method combines the expertise of clinicians with advanced sensor-based technology to capture and analyze subtle movement variations often unobserved by traditional subjective methods. Current approaches center on either body postures or muscle strength, which lack more sophisticated analysis features of muscle activation and coordination, thereby hindering analysis efficacy in deep rehabilitation feature exploration. To address this issue, we present a multimodal network algorithm that integrates surface electromyography (sEMG) and stress distribution signals. The algorithm considers the physical knowledge a priori to interpret the current rehabilitation stage and efficiently handles temporal dynamics arising from diverse user profiles in an online setting. Besides, we verified the performance of this model using a learned-nonuse phenomenon assessment task in 24 subjects, achieving an accuracy of 94.7%. Our results surpass those of conventional feature-based, distance-based, and ensemble baseline models, highlighting the advantages of incorporating multimodal information rather than relying solely on unimodal data. Moreover, the proposed model presents a network design solution for rehabilitation physical training that requires deep bioinformatic features and can potentially assist real-time and home-based physical training work.

Human activity recognition (HAR) is now a powerful in-home fitness assistive technology. This article presents a wearables-based multisensor system (WMS), which not only supports conventional functionalities, such as motion evaluation based on multidimensional information about the user’s body (movement speed, angle, muscle states, etc.), but also provides advanced services, including assessing training fatigue and providing real time, elastic, and professional training advice. The proposed WMS is experimentally validated by yielding 90.11% accuracy of motion evaluation with 36% and 23% improvement of fitness effect on bicep girth and muscular endurance, indicating its feasibility to prompt the development of HAR in the in-home fitness training domain.

In recent years, human activity recognition (HAR) technologies in e-health have triggered broad interest. In literature, mainstream works focus on the body's spatial information (i.e. postures) which lacks the interpretation of key bioinformatics associated with movements, limiting the use in applications requiring comprehensively evaluating motion tasks' correctness. To address the issue, in this article, a Wearables-based Multi-column Neural Network (WMNN) for HAR based on multi-sensor fusion and deep learning is presented. Here, the Tai Chi Eight Methods were utilized as an example as in which both postures and muscle activity strengths are significant. The research work was validated by recruiting 14 subjects in total, and we experimentally show 96.9% and 92.5% accuracy for training and testing, for a total of 144 postures and corresponding muscle activities. The method is then provided with a human-machine interface (HMI), which returns users with motion suggestions (i.e. postures and muscle strength). The report demonstrates that the proposed HAR technique can enhance users' self-training efficiency, potentially promoting the development of the HAR area.

The combination of teleoperating robots and the Internet of Things (IoT) could be employed in many areas, including remote nursing and semi-mechanical control. However, it is known that subjects can quickly encounter physical and mental fatigue, which can potentially lower accuracy in teleoperation. To address this issue, this article presents a closed-loop teleoperation system based on multisensory fusion, visual, and haptic feedback within the IoT framework. Various sensors for electromyography, inertial measurement unit, and mechanical hand control system are deployed to obtain body signals from participants, subsequently processed by artificial intelligence methods. Resistive sensors are installed at the robotic hand side to learn the contact force between the robot hand and the grasped object. To allow the user to understand the contact force levels, a haptic interface is equipped to provide three-level mechanical vibrations. To help users identify and trace different objects, a region convolution neural network for detecting 100 different categories of things is constructed. In addition, the real-time control (delay$\mathbf {< } 110$ms) is offered by a tri-layered IoT architecture. The feasibility of the proposed technique is validated by a grasping task carried out by 20 volunteers. During experiments, it is observed that merely 10.8 s is averagely needed for participants in finishing the task, with a high average success rate of 97%. Besides, NASA-TLX questionnaire and maximum voluntary contraction test report that users suffer light mental and physical fatigue when the proposed technique is used.