Human Action Recognition (HAR) offers a wide range of applications, including smart home, smart health, entertainment, security, and surveillance. Traditional vision-based HAR systems face significant limitations due to privacy concerns, lighting dependency, and poor performance in complex environments. Millimeter-wave radar-based activity recognition systems have attracted considerable attention due to their superior sensing capabilities, device-free deployment, privacy preservation, and robustness to environmental variations. However, existing approaches struggle with the inherent sparsity and noise in mmWave radar data, particularly for diverse activity categories spanning from full-body movements to subtle localized gestures. This study proposes mm-ARnet, a comprehensive millimeter-wave point-cloud-based framework for recognizing 16 diverse human activities across three distinct behavioral categories: full-body movements, posture transitions, and localized body movements. Our approach leverages 4D point cloud sequences and introduces a multi-frame fusion with stochastic sampling strategy to enhance point cloud density and mitigate sparsity effects. The core innovation lies in our lightweight TCN+Bi-LSTM temporal modeling pipeline integrated with a novel Temporal Pattern Attention (TPA) mechanism. Extensive experiments conducted across three real-world scenarios with 10 participants demonstrate that mm-ARnet achieves 97.42% accuracy, outperforming state-of-the-art methods while maintaining superior temporal performance.

Human action recognition in classrooms has recently become a research hotspot. Traditional solutions usually rely on sensors or computer vision methods. However, these methods have some disadvantages, such as difficulty in deployment, susceptibility to ambient light, and privacy and security issues. This paper proposes RFAR, a contactless method for classroom action recognition. This method utilizes an RFID tag placed on the desktop to capture various actions and subsequently evaluate the student's learning status. To enhance the reliability of singletag identification, fused data consisting of two or three types of data sequences (RSSI, phase, and Doppler shift) are incorporated. Furthermore, a dynamic antenna system is utilized to identify the optimal angle for tag-antenna alignment. Notably, the single-tagper-person design eliminates severe interference among multiple tags and simplifies device deployment in multi-person scenarios. This method is proposed based on COTS RFID devices and shows high robustness across different environments and equipment. Experimental results show a recognition accuracy of 93.9% in single-person scenarios and 81.5% in five-person scenarios.

The median (MED) is a crucial statistic for measuring the central tendency. However, exact MED computation remains costly, with even state-of-the-art (SOTA) algorithms failing to meet (near) real-time processing demands. While approximate MED algorithm has arisen as a promising candidate, existing approaches ignore the potential opportunity of spatiotemporal similarity within the application and fail to provide applicationspecific trade-offs between execution time and accuracy. Our goal is to design an enhanced approximate MED algorithm STREAM, which is capable of exploiting the spatiotemporal similarity to achieve bucket reuse and establish a tunable-grained bucket mechanism to meet the accuracy of application-specific requirements. Experimental results show that while maintaining nearly identical accuracy, STREAM outperforms the SOTA approximate methods DDSketch (up to $10 \times 4.7 \times$ on average) and KLL (up to $71.2 \times 10.1 \times$ on average).

Sparse general matrix-matrix multiplication (SpGEMM) is a core primitive for numerous scientific applications. Traditional hash-based approaches fail to strike a balance between reducing hash collisions and efficiently utilizing fast shared memory, which significantly undermines the performance of executing SpGEMM on GPUs. To address this issue, this paper introduces a novel accumulator design that achieves high shared memory utilization on modern GPUs. For the proposed high shared memory utilization algorithm, i.e., HSMU-SpGEMM1, we further optimize different symbolic stages. Our evaluations with four state-of-the-art hash-based SpGEMM libraries (Nsparse, spECK, OpSparse, and NVIDIA’s cuSPARSE) on three NVIDIA GPUs (Ampere, Ada Lovelace, Turing) demonstrate significant performance benefits from HSMU-SpGEMM.1HSMU-SpGEMM is available at https://github.com/wuminqaq/HSMUSpGEMM

Information extraction is pivotal in natural language processing, where the goal is to convert unstructured text into structured information. A significant challenge in this domain is the diversity and specific needs of various processing tasks. Traditional approaches typically utilize separate frameworks for different information extraction tasks, such as named entity recognition and relationship extraction, which hampers their uniformity and scalability. In this study, this study introduce a Universal Information Extraction (UIE) framework combined with a cue learning strategy, significantly improving the efficiency and accuracy of extracting mine hoist fault data. Initially, domain-specific data is manually labeled to fine-tune the model, and the accuracy is further enhanced by constructing negative examples during this fine-tuning process. The model then focuses on faults using the Structured Extraction Language (SEL) and a schema-based prompt syntax, the Structural Schema Instructor (SSI), which targets and extracts key information from the fault data to meet specific domain requirements. Experimental results show that UIE substantially improves the processing efficiency and the F1 accuracy of the extracted mine hoist fault data, with the fine-tuned F1 score increasing from 23.59% to 92.51%.

With the increasing scale of cloud computing applications of next-generation embedded systems, a major challenge that domain scientists are facing is how to efficiently store and analyze the vast volume of output data. Compression can reduce the amount of data that needs to be transferred and stored. However, most of the large datasets are in floating-point format, which exhibits high entropy. As a result, existing lossless compressors cannot provide enough performance for such applications. To address this problem, we propose a total variation reduction method for improving the compression ratio of lossless compressors (namely, FPC + and FPZIP + ), which employs a median-based hyperplane to precondition the data. In particular, we first try to exploit the space-filling curve (SFC), a well-known technique to preserve data locality for a multi-dimensional dataset. We show and explain why a raw SFC, such as Hilbert and Z-order curves, cannot improve the compression ratio. Then, we explore the opportunity and theoretical feasibility of the proposed total variation reduction-based algorithm. The experiment results show the effectiveness of the proposed method. The compression ratios are improved up to 48.2% (20.6% on average) for FPZIP and 42.4% (18.4% on average) for FPC. Moreover, through observing the time composition of the proposed method, it is found that the median finding holds a high percentage of the execution time. Hence, we further introduce an approximate median finding algorithm, providing a linear-time overhead reduction scheme. The experiment results clearly demonstrate that this algorithm reduces execution time by an average of 56.7% and 40.7% compared to FPC + and FPZIP + , respectively.

Data compression can efficiently reduce the memory and persistence storage cost, which is highly desirable in modern computing systems, such as enterprise, cloud, and High-Performance Computing (HPC) environments. However, the main challenges of existing data compressors are the insufficient compression ratio and low throughput. This paper focuses on improving the compression ratio of state-of-the-art lossy compression algorithms from the view of applications. Besides, we also use the characteristics of the applications to reduce the runtime overhead. To this end, we explore the idea with Adaptive Mesh Refinement (AMR), which is widely adopted as a computational technique to reduce the amount of computation and memory required in scientific simulations. We propose Level Associated Mapping-based Preconditioning (LAMP) to improve the storage efficiency of AMR applications. The main idea is twofold. First, we utilize the high similarities among the adjacent AMR levels to precondition the data prior to compression. Second, AMR has a unique characteristic of grid structures. We utilize grid structures to rebuild a level associated mapping table, which significantly reduces the runtime overhead of LAMP. Thanks to the optimization techniques of General Matrix Multiplication (GEMM), we further accelerate the process of rebuilding AMR hierarchy for LAMP. Besides, we also block multiple adjacent coordinates within a box and further improve cache locality. The experimental results show that the compression ratios of LAMP are improved up to 63.8% compared to directly compressing the data.

Tibetan is a low-resource language with few existing electronic reference materials. The goal of Tibetan sentence boundary disambiguation (SBD) is to segment long text into sentences, and it is the foundation for downstream tasks corpora building. This study implemented the Tibetan SBD at the syllable level to avoid word segmentation (WS) errors affecting the accuracy of SBD. Specifically, the attention mechanism is introduced based on a recurrent neural network (RNN) to study Tibetan SBD. The primary objective is to determine, using a trained model, whether the shad contained in Tibetan text is the ending of the sentence, and implement experiments on syllable embedding and component embedding to measure the model's performance. The highest accuracy for Tibetan syllable embedding and component embedding is 96.23% and 95.40 %, respectively, and the F1 score reaches 96.23% and 95.37%, respectively. The experimental results demonstrate that the proposed method can achieve better results than the established rule-based and statistical methods without considering various syntactic and part-of-speech (POS) tagging rules. German and English data from the Europarl corpus and Thai data from the IWSLT2015 corpus are validated to prove the models’ reliability and generalizability. The results demonstrate that this method is efficient not only for low-resource languages but also for high-resource languages. More importantly, we can formally apply the experimental results of this study to the research of downstream tasks, such as machine translation and automatic summarization.

Tibetan word segmentation and POS tagging are the primary tasks of Tibetan natural language processing. Most of existing methods of Tibetan word segmentation and POS tagging are based on rules and statistics, which need manual construction of features. In addition, the joint mode has shown stronger capabilities for word segmentation and POS tagging and have received great interests. In this paper, we propose Bi-LSTM+IDCNN+CRF structures, a simple yet effective end-to-end neural network model, for joint Tibetan word segmentation and POS tagging. We conduct step-by-step and joint experiments on the Tibetan datasets. The results demonstrate that the performance of the Bi-LSTM+IDCNN+CRF model is the best regardless of the step-by-step or joint mode. We obtain state-of-the-art performance in the joint tagging mode. The F1 score of the word segmentation task reached 92.31%, and the F1 score of the POS tagging task reached 81.26%.

A modified Levenberg-Marquardt (LM) method is proposed to improve the waveform-decomposition efficiency of light detection and ranging (LiDAR). The conventional constant-model-based LM fitting algorithm is subsequently modified using two proposed models: the linear model and exponential model. By revising the update coefficient of the damping term to make it consistent with the variation of residual error, the magnitude of oscillation is effectively reduced to provide better convergence. The models were experimentally verified using observed data acquired by our experimental large-footprint LiDAR system. The results indicate that the two modified LM-based algorithms provide better performance in terms of convergence speed and iteration efficiency for waveform decomposition in comparison with the traditional algorithm. Most prominently, the exponential LM algorithm provides 69% maximum improvement in convergence speed and 103% in acceptable iteration efficiency in comparison with the traditional algorithm.