Short text stream clustering is a challenging task since 1) the limited words in short texts lead to text sparsity, making it difficult to capture the comprehensive semantic information of texts, and 2) the topic evolution leads to catastrophic forgetting of the learning model, causing a degradation in clustering performance. To address these issues, we propose a semi-supervised short text stream clustering method with Dual-View semantic representation and Cluster-Complement learning (DVCC). Specifically, to address insufficient semantic information caused by text sparsity, we introduce dual-view semantic representation, consisting of the word-level semantic view and the sentence-level semantic view. We model the word-level semantic view using the Dirichlet Process Multinomial Mixture (DPMM) model and generate the sentence-level semantic view by an encoder. These two views are then integrated into a unified model for online clustering via a Bernoulli random variable. To address catastrophic forgetting caused by topic evolution, we introduce cluster-complement learning to update the encoder offline periodically. We initially use a small amount of labeled data to initialize the parameters of the encoder, and then use a large amount of unlabeled texts and their pseudo-labels as training data for subsequent updates. The proposed cluster-complement learning supplements the training data with cluster centroids, ensuring the centroids of all active clusters appear in each batch of data to maintain knowledge of old topics. Finally, we validate the effectiveness of our method by conducting extensive experiments on six benchmark datasets. The experimental results show that DVCC outperforms existing state-of-the-art methods. The code is available athttps://github.com/yaru-Zhangz/DVCC.

Weakly-supervised temporal sentence grounding remains challenging when learning to temporally locate event boundaries in a video related to the given query. Conventional methods that rely on global query supervision suffer from key limitations (e.g. insufficient interactions between local video-query representations). To tackle it, in this paper, we propose the ConceptNet which achieves fine-grained alignments by leveraging concept-centric learning. Specifically, we extract the essential concepts (i.e., verbs and nouns) and design two corresponding networks: Temporal-Dynamic Network (TDNet) and Visual-Semantic Network (VSNet). The TDNet introduces a prompt-guided autoregressive task aimed at facilitating the learning of temporal dependencies, with the objective of enhancing the model more sensitive to event progression rather than static scenes. Besides, the VSNet is designed to answer the masked query templates from batch-wise concept pools for semantic alignments. Extensive evaluations on Charades-STA and ActivityNet Captions show the superiority of our ConceptNet when compared to previous state-of-the-arts. The code is available at https://github.com/rubyrosecraft/ConceptNet.

MOTIVATION: The utilization of single-cell bisulfite sequencing (scBS-seq) methods allows for precise analysis of DNA methylation patterns at the individual cell level, enabling the identification of rare populations, revealing cell-specific epigenetic changes, and improving differential methylation analysis. Nonetheless, the presence of sparse data and an overabundance of zeros and ones, attributed to limited sequencing depth and coverage, frequently results in reduced precision accuracy during the process of differential methylation detection using scBS-seq. Consequently, there is a pressing demand for an innovative differential methylation analysis approach that effectively tackles these data characteristics and enhances recognition accuracy. RESULTS: We propose a novel beta mixture approach called scDMV for analyzing methylation differences in single-cell bisulfite sequencing data, which effectively handles excess zeros and ones and accommodates low-input sequencing. Our extensive simulation studies demonstrate that the scDMV approach outperforms several alternative methods in terms of sensitivity, precision, and controlling the false positive rate. Moreover, in real data applications, we observe that scDMV exhibits higher precision and sensitivity in identifying differentially methylated regions, even with low-input samples. In addition, scDMV reveals important information for GO enrichment analysis with single-cell whole-genome sequencing data that are often overlooked by other methods. AVAILABILITY AND IMPLEMENTATION: The scDMV method, along with a comprehensive tutorial, can be accessed as an R package on the following GitHub repository: https://github.com/PLX-m/scDMV.

With the increasing popularity of intelligent surveillance systems, abnormal behavior detection of human beings based on computer vision is attracting more attention. It aims to classify and locate the abnormal behaviors and coordinates of human beings, respectively, and is a fundamental technology for intelligent security. Existing approaches mainly focus on exploring abnormal behavior features through object detectors. However, in office scenarios, almost all abnormal behaviors are closely associated with the fine-grained feature around the nose, wrist, elbow, and other human joint points regions. Detectors for generic objects cannot adequately capture such differences between abnormal behaviors, resulting in sub-optimal performance. In this paper, we focus on human joints and take one step further to enable effective behavior characteristics learning in office scenarios. In particular, we propose a novel Adaptive Joints Enhancement Network (AJENet), which includes two closely-related components, Joints Predict block (JP) and Adaptive Key Joints Enhancement block (AKJE). JP block is used to predict the human joints and facilitates the feature learning around them implicitly. By inputting the features around joints, the AKJE block enhances the feature representations of key joints according to the abnormal behavior characteristics adaptively. Experimental results demonstrate that our method outperforms other state-of-the-art methods on the collected real office scenario Office Behavior Dataset. Besides, to verify the generalization capabilities and potential of AJENet, we construct comparisons on another generic dataset PASCAL VOC 2012 Action.

Food authenticity is a significant concern in the meat industry, demanding effective detection methods.This study explores the use of multispectral imaging (MSI) and deep learning for meat adulteration detection.We evaluate different deep learning models using transfer learning and preprocessing techniques in a multi-level adulteration classification task.In addition, we propose a novel approach called one-band mixed augmentation for band selection in MSI data, which outperforms traditional reflectance-based feature selection and enhances model robustness.Furthermore, employing the ninecrop approach for dataset augmentation improved the accuracy from 0.63 to 0.74 for DenseNet201 model without transfer learning.This research contributes to advancing food safety assessment practices and provides insights into the application of deep learning for preventing food adulteration.The proposed one-band mixed augmentation approach offers a novel strategy for handling band selection challenges in MSI data analysis.

The proliferation of single-cell multimodal sequencing technologies has enabled us to understand cellular heterogeneity with multiple views, providing novel and actionable biological insights into the disease-driving mechanisms. Here, we propose a comprehensive end-to-end single-cell multimodal analysis framework named Deep Parametric Inference (DPI). DPI transforms single-cell multimodal data into a multimodal parameter space by inferring individual modal parameters. Analysis of cord blood mononuclear cells (CBMC) reveals that the multimodal parameter space can characterize the heterogeneity of cells more comprehensively than individual modalities. Furthermore, comparisons with the state-of-the-art methods on multiple datasets show that DPI has superior performance. Additionally, DPI can reference and query cell types without batch effects. As a result, DPI can successfully analyze the progression of COVID-19 disease in peripheral blood mononuclear cells (PBMC). Notably, we further propose a cell state vector field and analyze the transformation pattern of bone marrow cells (BMC) states. In conclusion, DPI is a powerful single-cell multimodal analysis framework that can provide new biological insights into biomedical researchers. The python packages, datasets and user-friendly manuals of DPI are freely available at https://github.com/studentiz/dpi.

SUMMARY: Fungi form a large and heterogeneous group of eukaryotic organisms with diverse ecological niches. The high importance of fungi contrasts with our limited understanding of fungal lifestyle and adaptability to environment. Over the last decade, the high-throughput sequencing technology produced tremendous RNA-sequencing (RNA-seq) data. However, there is no comprehensive database for mycologists to conveniently explore fungal gene expression and alternative splicing. Here, we have developed FungiExp, an online database including 35 821 curated RNA-seq samples derived from 220 fungal species, together with gene expression and alternative splicing profiles. It allows users to query and visualize gene expression and alternative splicing in the collected RNA-seq samples. Furthermore, FungiExp contains several online analysis tools, such as differential/specific, co-expression network and cross-species gene expression conservation analysis. Through these tools, users can obtain new insights by re-analyzing public RNA-seq data or upload personal data to co-analyze with public RNA-seq data. AVAILABILITY AND IMPLEMENTATION: The FungiExp is freely available at https://bioinfo.njau.edu.cn/fungiExp. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: At present, a fundamental challenge in single-cell RNA-sequencing data analysis is functional interpretation and annotation of cell clusters. Biological pathways in distinct cell types have different activation patterns, which facilitates the understanding of cell functions using single-cell transcriptomics. However, no effective web tool has been implemented for single-cell transcriptome data analysis based on prior biological pathway knowledge. RESULTS: Here, we present scTPA, a web-based platform for pathway-based analysis of single-cell RNA-seq data in human and mouse. scTPA incorporates four widely-used gene set enrichment methods to estimate the pathway activation scores of single cells based on a collection of available biological pathways with different functional and taxonomic classifications. The clustering analysis and cell-type-specific activation pathway identification were provided for the functional interpretation of cell types from a pathway-oriented perspective. An intuitive interface allows users to conveniently visualize and download single-cell pathway signatures. Overall, scTPA is a comprehensive tool for the identification of pathway activation signatures for the analysis of single cell heterogeneity. AVAILABILITY AND IMPLEMENTATION: http://sctpa.bio-data.cn/sctpa. CONTACT: [email protected] or [email protected] or [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In the stereo matching networks based on deep learning, current cost aggregation networks lack the means to aggregate cost volume to the utmost extent. Therefore, different from the standard encoder-decoder structures, we propose an attention aggregation encoder-decoder network framework for stereo matching that contains three modules. Specifically, we design a sub-branch and cross-stage aggregation encoding module, which aggregate context information of different sub-branches and cross-stages to achieve the mutual utilization of different deep cost volumes. Meanwhile, we introduce a three-dimensional attention recoding module to obtain the robust discriminative cost volume through recalibrating the high-level semantic information of the sub-branches. In addition, we construct a stepwise aggregation decoding module to decode the cost volume via the stepwise fusion upsampling strategy, which further enhances the learning ability of the network model. The experimental results on Scene Flow and KITTI benchmark datasets show that the proposed network framework is superior to other similar methods in aggregating information.

In this article, defects in the production process of silicon photovoltaic (Si-PV) cells are urgently needed to be detected due to their serious impact on the normal generation of PV system. In view of the shortcomings, such as low-defect efficiency, few detection data, and high detection error rate in the existing industrial production line, the main research purpose of this article is to complete an intelligent classification method for efficient and innovative defect detection for Si-PV cells and modules. The purpose is to improve the detection efficiency of Si-PV cell, to ensure the safety and reliability of Si-PV cell production process, to achieve large number of Si-PV cell defects detection and classification. First, the eddy current thermography system of Si-PV cells is established. Second, principal component analysis, independent component analysis, and nonnegative matrix factorization algorithms are compared for thermography sequences processing. Third, LeNet-5, VGG-16, and GoogleNet models are compared for Si-PV cell defects classification. Finally, the results show that the proposed method have successful application in Si-PV cell defects detection and classification.

Several studies have shown that the phone number and call behavior generated by a phone call reveal the type of phone call. By analyzing the phone number rules and call behavior patterns, we can recognize the fraudulent phone call. The success of this recognition heavily depends on the particular set of features that are used to construct the classifier. Since these features are human-labor engineered, any change introduced to the telephone fraud can render these carefully constructed features ineffective. In this paper, we show that we can automate the feature engineering process and, thus, automatically recognize the fraudulent phone call by applying our proposed novel approach based on deep learning. We design and construct a new classifier based on Call Detail Records (CDR) for fraudulent phone call recognition and find that the performance achieved by our deep learning-based approach outperforms competing methods. Experimental results demonstrate the effectiveness of the proposed approach. Specifically, in our accuracy evaluation, the obtained accuracy exceeds 99%, and the most performant deep learning model is 4.7% more accurate than the state-of-the-art recognition model on average. Furthermore, we show that our deep learning approach is very stable in real-world environments, and the implicit features automatically learned by our approach are far more resilient to dynamic changes of a fraudulent phone number and its call behavior over time. We conclude that the ability to automatically construct the most relevant phone number features and call behavior features and perform accurate fraudulent phone call recognition makes our deep learning-based approach a precise, efficient, and robust technique for fraudulent phone call recognition.