As a vital technology in Cyber-Physical Social Intelligence (CPSI), Multi-Object-Tracking (MOT) can support comprehensive perception and analysis of the physical environment and social virtual space, promoting an in-depth understanding of human behavior, object movement, and social interaction. Most MOT methods often adopt simple interpolation or prediction strategies when dealing with temporarily lost targets, but ignore the comprehensive consideration of the state of the target before its reappearance. This approach may lead to an incomplete understanding of the target's behavior and dynamics, which affects the accuracy and depth of the comprehensive understanding of social and physical space interactions in the real world. To improve it, we propose an online multi-object tracking method based on Record Confidence and Hierarchical Association (RCHA), which is represented as RCHA-Track. The Kalman filter combined with an Enhanced Correlation Coefficient (ECC) provides more accurate motion prediction under the influence of camera motion. The record confidence is designed to evaluate the loss status of the unseen object and refine the tracking trajectory. The normally tracked targets and the temporarily lost targets are combined to perform a hierarchical association based on the number of lost frames to achieve more accurate data associations. Compared with the latest ByteTrack, RCHA-Track improves MOTA, IDF1, and HOTA by 1.7%, 1.6%, and 1.3% on the benchmark dataset MOT17, and 1.3%, 2.1%, and 2.0% on MOT20, respectively, achieving state-of-the-art performance. Extensive ablation experiments demonstrate the effectiveness of each key module in the proposed RCHA-Track.

Predicting mortality risk in the Intensive Care Unit (ICU) using Electronic Medical Records (EMR) is crucial for identifying patients in need of immediate attention. However, the incompleteness and the variability of EMR features for each patient make mortality prediction challenging. This study proposes a multimodal representation learning framework based on a novel personalized graph-based fusion approach to address these challenges. The proposed approach involves constructing patient-specific modality aggregation graphs to provide information about the features associated with each patient from incomplete multimodal data, enabling the effective and explainable fusion of the incomplete features. Modality-specific encoders are employed to encode each modality feature separately. To tackle the variability and incompleteness of input features among patients, a novel personalized graph-based fusion method is proposed to fuse patient-specific multimodal feature representations based on the constructed modality aggregation graphs. Furthermore, a MultiModal Gated Contrastive Representation Learning (MMGCRL) method is proposed to facilitate capturing adequate complementary information from multimodal representations and improve model performance. We evaluate the proposed framework using the large-scale ICU dataset, MIMIC-III. Experimental results demonstrate its effectiveness in mortality prediction, outperforming several state-of-the-art methods.

Zero-Shot object Detection (ZSD), one of the most challenging problems in the field of object detection, aims to accurately identify new categories that are not encountered during training. Recent advancements in deep learning and increased computational power have led to significant improvements in object detection systems, achieving high recognition accuracy on benchmark datasets. However, these systems remain limited in real-world applications due to the scarcity of labeled training samples, making it difficult to detect unseen classes. To address this, researchers have explored various approaches, yielding promising progress. This article provides a comprehensive review of the current state of ZSD, distinguishing four related methods—zero-shot, open-vocabulary, open-set, and open-world approaches—based on task objectives and data usage. We highlight representative methods, discuss the technical challenges within each framework, and summarize the commonly used evaluation metrics, benchmark datasets, and experimental results. Our review aims to offer readers a clear overview of the latest developments and performance trends in ZSD.

In today's rapidly evolving internet landscape, prominent companies across various industries face increasingly complex business operations, leading to significant cluster-scale growth. However, this growth brings about challenges in cluster management and the inefficient utilization of vast amounts of data due to its low value density. This paper, based on the large-scale cluster virtualization and monitoring system of the data center of the Bureau of Geophysical Prospecting (BGP), utilizes time series data of host resources from the monitoring system's time series database to propose a multivariate multi-step time series forecasting model, MUL-CNN-BiGRU-Attention, for forecasting CPU load on virtual cluster hosts. The model undergoes extensive offline training using a large volume of time series data, followed by deployment using TensorFlow Serving. Recent small-batch data are employed for fine-tuning model parameters to better adapt to current data patterns. Comparative experiments are conducted between the proposed model and other baseline models, demonstrating notable improvements in Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> metrics by up to 35.2%, 56.1%, 32.5%, and 10.3%, respectively. Additionally, ablation experiments are designed to investigate the impact of different factors on the performance of the forecasting model, providing valuable insights for parameter optimization based on experimental results.

Using skeletal information to model and recognize human actions is currently a hot research subject in the realm of Human Action Recognition (HAR). Graph Convolutional Networks (GCN) have gained popularity in this discipline due to their capacity to efficiently process graph-structured data. However, it is challenging for current models to handle distant dependencies that commonly exist between human skeleton nodes, which hinders the development of algorithms in related fields. To solve these problems, the Lightweight Multiscale Spatio-Temporal Graph Convolutional Network (LMSTGCN) is proposed. Firstly, the Lightweight Multiscale Spatial Graph Convolutional Network (LMSGCN) is constructed to capture the information in various hierarchies, and multiple inner connections between skeleton joints are captured by dividing the input features into a number of subsets along the channel direction. Secondly, the dilated convolution is incorporated into the temporal convolution to construct Lightweight Multiscale Temporal Convolutional Network (LMTCN), which allows to obtain a wider receptive field while keeping the size of the convolution kernel unchanged. Thirdly, the Spatio-Temporal Location Attention (STLAtt) module is used to identify the most informative joints in the sequence of skeletal information at a specific frame, hence improving the model's ability to extract features and recognize actions. Finally, multi-stream data fusion input structure is used to enhance the input data and expand the feature information. Experiments on three public datasets illustrate the effectiveness of the proposed network.

The effectiveness of recommendation systems heavily relies on accurately predicting user ratings for items based on user preferences and item attributes derived from ratings and reviews. However, the increasing presence of fake user data in these ratings and reviews poses significant challenges, hindering feature extraction, diminishing rating prediction accuracy, and eroding user trust in the system. To tackle this issue, we propose a robust rating prediction model for recommendation systems that integrates fake user detection and multi-layer feature fusion. Our model utilizes a GraphSAGE-based submodel to filter out fake user data from rating data and review texts. To strengthen fake user detection, we enhance GraphSAGE by selecting aggregation neighbors based on the collusion fraud degree among users, and employ an attention mechanism to weigh the contribution of each neighbor during representation aggregation. Furthermore, we introduce a multi-layer feature fusion submodel to integrate diverse features extracted from the filtered ratings and reviews. For deep feature extraction from review texts, we implement a temporal attention mechanism to analyze the relevance of reviews over time. For shallow feature extraction from rating data, we incorporate trust evaluation mechanism and cloud model to assess the influence of trusted neighbors' ratings. In our evaluation, we compare our model against six baseline models for fake user detection and four rating prediction models across five datasets. The results demonstrate that our model exhibits significant performance advantages in both fake user detection and rating prediction.

The advent of Large Language Models (LLMs) has sparked considerable interest in the medical image domain, as they can generalize to multiple tasks and offer outstanding performance. While LLMs achieve promising results, there is currently a lack of a comprehensive summary of medical images, making it challenging for researchers to understand the progress within this domain. To fill this gap, we make the first attempt to present a comprehensive survey for LLM on medical images. In addition, to better summarize the current progress comprehensively, we further introduce a novel x-stage tuning paradigm for summarization, including zero-stage tuning, one-stage tuning, and multi-stage tuning, offering a unified perspective on LLMs for medical images. Finally, we discuss challenges and future directions in this domain, aiming to spur more breakthroughs in the future. We hope this work can pave the way for the broad application of LLMs in medical images and provide a valuable resource for this domain.

Non-negative Matrix Factorization (NMF) has been an ideal tool for machine learning. Non-negative Matrix Tri-Factorization (NMTF) is a generalization of NMF that incorporates a third non-negative factorization matrix, and has shown impressive clustering performance by imposing simultaneous orthogonality constraints on both sample and feature spaces. However, the performance of NMTF dramatically degrades if the data are contaminated with noises and outliers. Furthermore, the high-order geometric information is rarely considered. In this paper, a Robust NMTF with Dual Hyper-graph regularization (namely RDHNMTF) is introduced. Firstly, to enhance the robustness of NMTF, an improvement is made by utilizing the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$l_{2,1}$</tex>-norm to evaluate the reconstruction error. Secondly, a dual hyper-graph is established to uncover the higher-order inherent information within sample space and feature spaces for clustering. Furthermore, an alternating iteration algorithm is devised, and its convergence is thoroughly analyzed. Additionally, computational complexity is analyzed among comparison algorithms. The effectiveness of RDHNMTF is verified by benchmarking against ten cuttina-edae alaorithms across seven datasets corrupted with four types of noise.

The evolution of lung lesions can be assessed by examining multiple CT screenings, which needs to align two CT images accurately. In this study, we propose a multi-stride and multi-resolution affine registration network, called MMAR-net, for 3D affine registration of medical images, which works in an unsupervised way by optimizing the similarity loss. In order to extract more extensive image features, we use a multi-stride module to replace the conventional convolution module. Furthermore, we make use of the image features at multiple scales by dot product between two feature vectors, which could enhance the robustness of image representation. We conduct comprehensive comparison experiments between our model and the existing affine registration methods on two publicly available datasets, DIR-Lab and Learn2Reg, which are both relevant to lung CT image registration. Quantitative and qualitative comparison results demonstrate that our model outperforms existing single-step affine registration networks. Our method improves the key metric of dice similarity coefficient on DIR-Lab and Learn2Reg to 90.57% and 95.51%, respectively.

Recent years have witnessed exponential growth in both the volume and complexity of biomedical data generated, which encompasses a diverse array of sources, including genomic sequences, clinical imaging, electronic health records, and real-time biometric sensing. This deluge of data has surpassed our traditional methods' abilities to efficiently process, analyze, store, and interpret, posing substantial challenges across the biomedical and healthcare sectors. These challenges are not only technical, involving data management and computational efficiency but also extend to ethical, legal, and privacy concerns. Despite these challenges, significant opportunities exist for advancements in personalized medicine, disease biomarker prediction, and early diagnosis of diseases. We are pleased to present this special issue titled "Challenges and Opportunities in Biomedical Big Data Analysis: From Large Language Models to Clinical Applications", which showcases pioneering research and stateof-the-art review in this dynamic field. The rigorous peer-review process conducted by expert reviewers contributed to refining and improving the quality of the final manuscripts. In this issue, we have accepted ten insightful papers that cover three main thematic categories: language models and single-cell analysis, medical imaging and biomarker detection technologies, as well as genomics and big data analysis.