Outlier interpretation methods, based on outlying aspect mining, have been extensively utilized in diverse applications due to their effectiveness and interpretability. The primary objective of these methods is to identify an outlying feature subspace where a detected outlier deviates most significantly from the inliers. However, this subspace is typically not personalized and incomplete. In this article, we propose a novel convertible outlying aspect mining method named Mining convErtible ouTlying Aspect (META) to interpret the detected outlier. META not only identifies a personalized outlying feature subspace (i.e., where) that differentiates the detected outlier from inliers, but also quantifies the outlying direction and degree within this subspace (i.e., how), thereby offering actionable interpretative insights, rather than corrective actions, for understanding the outlier. Specifically, META defines a convertible outlying aspect with a convertible cost to convert the detected outlier into a converted instance, and employs a pretrained adversary to evaluate whether the instance is an inlier or not. Subsequently, we formulate an objective function that minimizes the convertible cost, ensures the successful conversion of the instance into an inlier, and minimizes the size of the outlying feature subspace. META leverages this objective function to learn an optimal convertible outlying aspect for the detected outlier. The optimal convertible outlying aspect provides the outlying feature subspace, outlying direction, outlying degree, and converted instance. Empirical results from experiments conducted on both real-world and synthetic datasets demonstrate that META outperforms state-of-the-art (SOTA) baselines.

Natural language transformation (NLT) tasks, such as machine translation (MT) and text style transfer (TST), require models to generate accurate and contextually appropriate outputs.However, existing approaches face significant challenges, including the computational costs of leveraging large pre-trained models and the limited generalization ability of finetuned smaller models.In this paper, we propose a novel framework that combines the flexibility of prompting with the cost-effectiveness of fine-tuning.Our method enhances smaller models by integrating In-Context Examples (ICE) from retrieval, enabling the model to better capture contextual information and align with userlevel preferences.We further improve performance through hierarchical contrastive learning and dynamic preference inference mechanisms.Experimental results demonstrate that our approach outperforms existing methods, such as Supervised Fine Tuning (SFT), Direct Preference Optimization (DPO), and Contrastive Preference Optimization (CPO), across both MT and TST tasks, providing a more efficient solution for resource-constrained environments.

Anomaly detection suffers from data imbalance, as anomalies are typically rare. Due to the smaller number of anomaly samples, models may overfit the features of the normal samples, resulting in poor performance when detecting anomalies. Historically, this issue has typically been addressed by undersampling normal data or oversampling anomalous data. Most methods have a significant drawback when oversampling the anomalous data: a lack of interpretability. Synthesized anomalous data often lack the background and contextual information of real data, making it difficult to interpret the reasons behind their synthesis. This lack of underlying reasons for synthesis can prevent the application of such methods in critical anomaly detection domains, such as medical diagnosis and autonomous driving. Therefore, we propose a Diffusion Counterfactual-Based Anomaly Detection (DCFAD) model to generate more interpretable anomalous data, addressing the issue of the scarcity of anomalous data. Through extensive experiments, the effectiveness of the proposed DCFAD model was validated on multiple datasets across two metrics.

Ensemble clustering learns a consensus result by integrating a set of base clustering results. Recently, anchor-based methods construct an anchor similarity matrix to represent the affinity relationships among samples, significantly improving computational efficiency. However, these methods struggle with fixed anchors generated by static anchor learning strategies, which lead to low-quality anchor similarity matrix and poor clustering accuracy. To address this issue, we propose a novel method named dynamic anchor-based ensemble clustering via hypergraph reconstruction (YACHT). Specifically, YACHT first transforms the base clustering results into a hypergraph and designs a novel hypergraph enhancement strategy to improve the reliability of the initial hypergraph. YACHT reconstructs the hypergraph through matrix factorization and introduces a mapping matrix to filter out redundant information, capturing a high-quality anchor similarity matrix. Then, YACHT attempts to incorporate the hypergraph into the optimization objective to achieve hypergraph updates. To ensure the accuracy of hypergraph updates, we impose a hypergraph regularizer and a local consensus information alignment term. The alignment term is implemented by minimizing the discrepancy between the label partition derived from the hypergraph regularizer and the local consensus information indicator matrix extracted from the base clustering results. Extensive experimental results demonstrate the outstanding performance of the proposed YACHT. The code is available at https://github.com/scu-kdde/YACHT.

Clustering, a fundamental task in machine learning and data mining, is essential for uncovering patterns by grouping data points with similar characteristics. Traditional methods struggle with nonlinear data structures, but kernel-based approaches alleviate this issue by mapping data to high dimensional spaces. Multiple Kernel Clustering (MKC) further improves clustering by automating kernel selection and integration. However, MKC faces challenges related to kernel graph quality, information loss during relax-and-discretization, neglect of balanced clustering constraints, and the trade-off between high clustering quality and balance. To address these challenges, we introduce Balanced Multiple Kernel Clustering (BMKC). BMKC utilizes local kernel reconstruction and advanced high-order diffusion techniques for comprehensive kernel graph learning. It directly learns a discrete partition matrix using a robust L1-induced local reconstruction criterion, eliminating the two step process. BMKC incorporates an automatic mechanism for trade-off control between clustering and balance, supported by a versatile optimization algorithm accommodating various balance regularization choices. Experimental validation demonstrates the superior performance of MKC on benchmarks data sets, showcasing its effectiveness. The code for our method is publicly available at https://github.com/ChenYan01TYUT/BMKC-ACM-MM-2025.

The rapid proliferation of multi-view data has necessitated robust and scalable clustering techniques capable of capturing complex, high-dimensional patterns. While Multi-view Bipartite Graph Clustering (MVBGC) has shown promising results, existing approaches often overlook that the generated bipartite graph is susceptible to disturbances from complex structures and noise. To address these challenges, we propose RTGD-MVC, a novel framework for Robust Tensor Learning with Graph Diffusion tailored for efficient and scalable multi-view graph clustering. RTGD-MVC integrates a graph diffusion mechanism to suppress noise propagation and employs cross-view diffusion to enhance global consistency while capturing complementary information across views. Additionally, a non-convex Tensor Exponential Norm (TEN) is introduced as a tighter surrogate for the tensor rank, enabling the learning of more discriminative and noise-robust representations. By embedding these components into a unified optimization model with linear computational complexity, RTGD-MVC achieves both theoretical efficiency and practical scalability. Extensive experiments on diverse benchmark datasets demonstrate that RTGD-MVC significantly outperforms state-of-the-art methods, highlighting its superior ability to capture intricate multi-view correlations and structural patterns.

Transformer-based large language models (LLMs) have demonstrated outstanding performance across diverse domains, particularly in the emerging pretrain-then-finetune paradigm. LoRA, a parameter-efficient fine-tuning method, is commonly used to adapt a base LLM to multiple downstream tasks. Further, LLM platforms enable developers to fine-tune multiple models and develop various domain-specific applications simultaneously. However, existing model parallelism schemes suffer from high communication overhead and inefficient GPU utilization. In this paper, we present mLoRA, a parallelism-efficient fine-tuning system designed for training multiple LoRA across GPUs and machines. mLoRA introduces a novel LoRA-aware pipeline parallelism scheme that efficiently pipelines LoRA adapters and their distinct fine-tuning stages across GPUs and machines, along with a new LoRA-efficient operator to enhance GPU utilization. Our extensive evaluation shows that mLoRA can significantly reduce average fine-tuning task completion time, e.g., by 30%, compared to state-of-the-art methods like FSDP. More importantly, mLoRA enables simultaneous fine-tuning of larger models, e.g., two Llama-2-13B models on four NVIDIA RTX A6000 48GB GPUs, which is not feasible for FSDP due to high memory requirements. Hence, mLoRA not only increases fine-tuning efficiency but also makes it more accessible on cost-effective GPUs.

In this article, we attempt to study a problem of identifying outliers from multiple dimensions, terming it as multidimensional anomaly detection. This task assumes that each sample corresponds to heterogeneous discriminative spaces, where each space characterizes distinct semantic information along one dimension. Consequently, the abnormalities exhibited by a sample will vary under these semantically different dimensions. In contrast to traditional anomaly detection, multidimensional anomaly detection offers a more holistic assessment of a sample's abnormalities. However, the heterogeneity of discriminative spaces leads to incomparability of the outputs from different dimensions which is the major difficulty in designing multidimensional anomaly detection methods. This article introduces a novel model, maximum margin multidimensional anomaly detection (ALOE), specifically tailored for multidimensional anomaly detection. ALOE constructs a convex optimization problem with nonlinear constraints. The primary objective is to simultaneously learn multiple decision boundaries, utilizing the maximum margin principle and covariance regularization, while distinguishing between outliers and normal samples under multiple dimensions by capturing the correlation among multiple dimensions. To obtain the optimal decision boundary under each dimension, we devise an alternating optimization method for this convex optimization problem. To validate the effectiveness of ALOE, we conduct extensive experiments on 12 real-world datasets, comparing its performance against 34 anomaly detection methods. The experimental results demonstrate the superior performance of ALOE.

Ensemble clustering learns more accurate consensus results from a set of weak base clustering results. This technique is more challenging than other clustering algorithms due to the base clustering result set's randomness and the inaccessibility of data features. Existing ensemble clustering methods rely on the Co-association (CA) matrix quality but lack the capability to handle missing connections in base clustering. Inspired by the neighborhood high-order and topological similarity theories, this paper proposes a topological ensemble model based on high-order information. Specifically, this paper compensates for missing connections by mining neighborhood high-order connection information in the CA matrix and learning optimal connections with adaptive weights. Afterward, the learned excellent connections are embedded into topology learning to capture the topology of the base clustering. Finally, we incorporate adaptive high-order connection representation and topology learning into a unified learning framework. To our knowledge, this is the first ensemble clustering work based on topological similarity and high-order connectivity relations. Extensive experiments on multiple datasets demonstrate the effectiveness of the proposed method. The source code of the proposed approach is available at https://github.com/ltyong/awec.

Improving the ability to predict protein function can potentially facilitate research in the fields of drug discovery and precision medicine. Technically, the properties of proteins are directly or indirectly reflected in their sequence and structure information, especially as the protein function is largely determined by its spatial properties. Existing approaches mostly focus on protein sequences or topological structures, while rarely exploiting the spatial properties and ignoring the relevance between sequence and structure information. Moreover, obtaining annotated data to improve protein function prediction is often time-consuming and costly. To this end, this work proposes a novel contrast-aware pre-training framework, called SCOP, for protein function prediction. We first design a simple yet effective encoder to integrate the protein topological and spatial features under the structure view. Then a convolutional neural network is utilized to learn the protein features under the sequence view. Finally, we pretrain SCOP by leveraging two types of auxiliary supervision to explore the relevance between these two views and thus extract informative representations to better predict protein function. Experimental results on four benchmark datasets and one self-built dataset demonstrate that SCOP provides more specific results, while using less pre-training data.

For a clinical prescription, clarifying the molecular mechanisms of actions (MMOAs) of the drug-disease interaction is helpful to optimize treatment, suggest possible side effects, and realize individualized treatment. Considering the relations among multiple biomedical entities, such as drugs, diseases, targets (genes), and pathways, what paths can be extracted connecting these biomedical entities to resemble the real mechanisms of a specific drug to a particular disease? Answering this question is crucial for understanding the underlying molecular mechanisms behind complex drug actions and identifying key pathways that can facilitate effective therapeutic interventions. In this paper, we propose an approach DDTExplainer that constructs a path-based graph neural network (GNN) explainer to mine the drug-disease therapeutic mechanisms. Technically, DDTExplainer transforms the drug-disease therapeutic mechanisms mining task into a GNN-based link prediction model explanation task. Firstly, a GNN-based drug-disease therapeutic prediction model is trained and joint-optimized with a translation-based graph embedding model. Secondly, mask learning is utilized to find the most prediction-influential edges and generate the path-based explanations with the shortest path algorithm. Finally, we assess the efficacy of DDTExplainer on a ground-truth dataset that consists of labeled entries describing the drug-disease therapeutic mechanisms. These labels are derived from well-established drug-target interactions, disease-target interactions, as well as target-pathway relationships, which were verified by wet experiments.

Dense retrieval (DR) has become a mainstream approach to information seeking, where a system is required to return relevant information to a user query. In real-life applications, typoed queries resulting from the users’ mistyping words or phonetic typing errors exist widely in search behaviors. Current dense retrievers experience a significant drop in retrieval effectiveness when they encounter typoed queries. Therefore, the search system requires the extra introduction of spell-checkers to deal with typos and then applies the DR model to perform robust matching. Herein, we argue that directly conducting the typos correction training would be beneficial to make an end-to-end retriever against misspellings. To this end, we propose a novel approach that can facilitate the incorporation of the spelling correction objective into the DR model using the encoder-decoder architecture. During typos correction training, we also develop a prompt-based augmentation technique to enhance the DR space alignment of the typoed query and its original query. Extensive experiments demonstrate that the effectiveness of our proposed end-to-end retriever significantly outperforms existing typos-aware training approaches and sophisticated training advanced retrievers. Our code is available at https://github.com/striver314/ToCoTR.

Multivariate time series anomaly detection (MT-SAD) plays a crucial role for the Internet of Things (IoT) systems. Various IoT systems rely on time series to monitor and identify anomalies, as well as to initiate remediation procedures. Existing contrastive learning methods for MTSAD have strong ability to learn invariant representations from augmented views. However, these methods rely on special data augmentation strategies and may not be suitable for certain time series. Besides, when abnormal points intensively occur in a segment, the similarity representations they learned from augmented views would make normal and abnormal points indistinguishable. To address these issues, we propose a novel Anchor-guided contrastive learning method with a general data augmentation strategy for multivariate time series Anomaly Detection (AgileAD). Specifically, this general strategy is a simple yet efficient data augmentation strategy to shuffle each univariate time series (i.e., metric) vertically. It can fit any time series to create different but correlated views. To improve the model discrimination in the presence of anomaly aggregation, AgileAD utilizes an anchor-guided contrastive structure module to learn the anchor representations of the raw input and capture temporal and intermetric dependencies of time series. Moreover, the triplet representation discrepancy is designed to promote the learning of shuffle invariant representations. It cooperates with the guided module to effectively distinguish between normal and abnormal points. Extensive experiments show that AgileAD achieves state-of-the-art results on multiple IoT benchmark datasets.

Recent advances in multi-kernel-based methods for outlier detection have positioned them as an attractive way to detect instances that are markedly different from the remaining data in a dataset. Currently, most outlier detection approaches based on multi-kernel learning are simply a convex combination of various kernels with handcrafted weights, meaning that these weights may not be suitable. Meanwhile, this combination of weights does not sufficiently consider the intrinsic correlations of instances when fusing different kernels. Thus, a key challenge is how to adaptively learn an appropriate combination of weights for capturing a new feature space in which outliers can be better detected than the original space. Simultaneously, it is still a burning issue to get the optimal combination of weights due to considerable computational cost and memory usage when the feature or instance size is large. In this paper, we propose a novel method forefficientadaptivemulti-kernel foroutlierdetection (EAMOD), which automatically learns the optimal weight for each training instance under different kernels using a non-negative function. In addition, we design a safe screening rule (SSR) for EAMOD to improve its training efficiency without any loss of accuracy. To the best of our knowledge, it is the first attempt to develop SSR for multi-kernel-based outlier detection methods. Extensive experiments show that EAMOD is effective and efficient.

Outlier detection methods based on distance measure have been used in numerous applications due to their effectiveness and interpretability. However, distances among instances heavily depend on the feature space in which they reside. For an outlier, distances from it to the normal instances may be extremely close in one feature space, failing to separate them from each other, while this situation is reversed in another space. Meanwhile, the distance measure is sensitive to a few “marginal instances” (i.e., normal instances located very close to outliers in the feature space) during the estimation of whether a test instance is an outlier or not. In this paper, we propose a robust multi-kernel nearest neighborhood (RMKN) method for outlier detection. Specifically, in the training phase, we only consider normal instances and transform them into a Polynomial kernel function weighted digraph to capture their geometric relationships in the original feature space. Then, we develop an objective function based on the weighted digraph to find a latent feature space via multi-kernel learning such that distances among normal instances in this latent feature space are as close as possible while preserving their original distributions. In the detecting phase, we design an outlying score based on the two-stage multi-kernel k-nearest nearest neighbors to detect outliers. Extensive experiments with ten datasets show that RMKN is effective and robust

Circular RNAs (circRNAs) play a critical role in gene regulation and association with diseases due to their specialized structure, which is formed as a closed loop structure during a non-canonical splicing process where the donor site back-spliced to an upstream acceptor site. As fundamental work to clarify their functions and mechanisms, a large number of computational methods for predicting circRNA formation have been proposed, among which, in particular, deep learning is utilized to capture relevant patterns from raw RNA sequences and model their interactions to facilitate prediction. However, these methods fail to fully utilize the important characteristics of back-splicing events, i.e., the positional information of the splice sites and the interaction features of its flanking sequences, for prediction. To this end, we hereby propose a novel approach called SIDE for predicting circRNA back-splicing events using only nucleotide sequences. Our model employs a dual encoder to capture global and interactive features of the sequence, and then a decoder designed by the contrastive learning to fuse out discriminative features improving the prediction of circRNAs formation. Empirical results on three real-world datasets have shown the effectiveness of SIDE. Our code is publicly available at https://github.com/scu-kdde/Bioinfo-SIDE-2023.

Predicting RNA-protein binding sites helps to explore the mechanisms of the interaction between RNA and proteins. Numerous deep learning methods have been applied to predict RNA-protein binding sites. Some of these methods use only sequence information for prediction which could lose information about the topology. And there may be a loss of important information if the secondary structure features are simply represented as one-hot matrices. Furthermore, existing deep learning methods are usually based on convolutional neural networks for feature extraction, which tend to focus on local features. As for the information of the whole sequence, existing methods usually ignore global features. Therefore, we propose a novel deep learning model called DARE for RNA-protein binding sites prediction using both sequence and secondary structure information of RNA. DARE employs the secondary structure feature extraction module to capture the features of the RNA secondary structure and learn the topological information. Therefore, we design a local feature extraction module and a global feature integration module to capture the whole information of RNA. Thus we can achieve the purpose of complementary information. Extensive experiments demonstrate that DARE outperforms baselines. Our analysis of the case study further confirm the effectiveness of DARE.

Drug-target interaction (DTI) prediction is of great importance for drug discovery and development. With the rapid development of biological and chemical technologies, computational methods for DTI prediction are becoming a promising strategy. However, there are few methods which explore solving the cold-start problem in DTI prediction scenarios due to most of existing methods require modeling under the existing interaction that can’t effectively capture information from new drugs and new targets which have few interactions in existing literature. In this paper, we propose a graph transformer method based on meta-learning named MGDTI to fill the gap. In particular, we employ drug-drug similarity and target-target similarity as additional information for network to mitigate the scarcity of interactions. Besides, we trained our model via meta-learning to be adaptive to cold-start tasks. Moreover, we introduced graph transformer to prevent over-smoothing by capturing long-range dependencies. Comparison results on the benchmark dataset demonstrate that our proposed MGDTI is effective in the DTI prediction.

Fraud detection aims to identify fraudsters from normal users. In graph environments, both fraudsters and normal users are modeled as nodes, while edges represent the connections between them. However, fraudulent nodes in the real world often camouflage themselves by establishing numerous fake connections with normal nodes, making them challenging to be identified. Existing fraud detection methods struggle to address this issue, they utilize graph neural networks to aggregate normal informations from normal neighbors, which leads to the smoothing of the fraudulent information. Furthermore, these methods exhibit poor generalization performance as they are unable to detect new fraudsters which not present in the training process. To overcome these limitations, this paper proposes GFAN, a novel model based on Graph Feature enhAncement Network. Specifically, GFAN introduces a specific semantic extraction module to screen and delete fake connections by evaluating the confidence level of edge presence. Additionally, GFAN provides a representation enhanced co-training module that highlights camouflaged fraudulent representations by training the small sphere and large margin support vector data description. Experimental results show that GFAN outperforms other competitive graph-based fraud detectors on public datasets. The GFAN code is available at: https://github.com/scu-kdde/OAM-GFAN-2023.

Underlying medical conditions, such as cancer, kidney disease and heart failure, are associated with a higher risk for severe COVID-19. Accurate classification of COVID-19 patients with underlying medical conditions is critical for personalized treatment decision and prognosis estimation. In this study, we propose an interpretable artificial intelligence model termed VDJMiner to mine the underlying medical conditions and predict the prognosis of COVID-19 patients according to their immune repertoires. In a cohort of more than 1400 COVID-19 patients, VDJMiner accurately identifies multiple underlying medical conditions, including cancers, chronic kidney disease, autoimmune disease, diabetes, congestive heart failure, coronary artery disease, asthma and chronic obstructive pulmonary disease, with an average area under the receiver operating characteristic curve (AUC) of 0.961. Meanwhile, in this same cohort, VDJMiner achieves an AUC of 0.922 in predicting severe COVID-19. Moreover, VDJMiner achieves an accuracy of 0.857 in predicting the response of COVID-19 patients to tocilizumab treatment on the leave-one-out test. Additionally, VDJMiner interpretively mines and scores V(D)J gene segments of the T-cell receptors that are associated with the disease. The identified associations between single-cell V(D)J gene segments and COVID-19 are highly consistent with previous studies. The source code of VDJMiner is publicly accessible at https://github.com/TencentAILabHealthcare/VDJMiner. The web server of VDJMiner is available at https://gene.ai.tencent.com/VDJMiner/.

Abstract Temporal heterogeneous graphs can model lots of complex systems in the real world, such as social networks and e-commerce applications, which are naturally time-varying and heterogeneous. As most existing graph representation learning methods cannot efficiently handle both of these characteristics, we propose a Transformer-like representation learning model, named THAN, to learn low-dimensional node embeddings preserving the topological structure features, heterogeneous semantics, and dynamic patterns of temporal heterogeneous graphs, simultaneously. Specifically, THAN first samples heterogeneous neighbors with temporal constraints and projects node features into the same vector space, then encodes time information and aggregates the neighborhood influence in different weights via type-aware self-attention. To capture long-term dependencies and evolutionary patterns, we design an optional memory module for storing and evolving dynamic node representations. Experiments on three real-world datasets demonstrate that THAN outperforms the state-of-the-arts in terms of effectiveness with respect to the temporal link prediction task.

Disease similarity analysis impacts significantly in pathogenesis revealing, treatment recommending, and disease-causing genes predicting. Previous works study the disease similarity based on the semantics obtaining from biomedical ontologies (e.g., disease ontology) or the function of disease-causing molecules. However, such methods almost focus on a single perspective for obtaining disease features, which may lead to biased results for similar disease detection. To address this issue, we propose a disease information network-based integrative approach named MISSION for detecting similar diseases. By leveraging the associations between diseases and other biomedical entities, the disease information network is established first. Then, the disease similarity features extracted from the aspects of disease taxonomy, attributes, literature, and annotations are integrated into the disease information network. Finally, the top-k similar disease query is performed based on the integrative disease information. The experiments conducted on real-world datasets demonstrate that MISSION is effective and useful in similar disease detection.

In dynamic ridesharing, the platform serves as the mediator by tailoring the assignment result between workers and riders with a focus on a certain objective. Existing studies generally focus on either one or two stakeholders when modelling the problem while the wellbeing of the other parties may be ignored or even undermined. For example, purely maximizing the total revenue of the ridesharing platform may cause the loss of riders and in turn lead to a low served rate, because those expensive orders will be processed in priority. In this paper, we for the first time study how to incorporate the willingness of all stakeholders (i.e., the platform, workers and riders). Given a set of workers and a set of rider requests, we aim to return the matchable worker-rider pairs in order to minimize theregret. Specifically, two types of regret are defined: (i) theserved rate regret, which refers to the rate of unserved requests, catering for the reputation and profit of the platform and workers; (ii) therevenue regret, which considers the portion of revenue loss from unserved riders, catering for the focus of workers and riders in the trip schedule. We prove the NP-hardness of this problem. To tackle this problem, we first propose a dynamic programming insertion algorithm to improve the efficiency of inserting a rider request into a trip schedule of a worker. Furthermore, two kinds of heuristic algorithms are devised to match rider requests with workers effectively. Comprehensive experiments on two real-world datasets verify the effectiveness, efficiency and scalability of our solutions in dealing with different supply-demand relationships in practice.

Gene similar aspects provide reliable explanation in understanding the biological roles and gene functions. As the volume of biomedical data expands, most of the current methods for similar explanation among genes are no longer applicable. Limited by information sources and search effciency, these methods cannot be flexible and effcient for the similarity analysis. We hereby propose a flexible method VENUS to analyze gene similar aspect among multiple genes on heterogeneous information networks, which constructed from public biomedicine databases and literature. VENUS infers the semantic and structural similarity of the query genes by gene community search. In this way, VENUS narrows the search space when searching information network within an acceptable time cost. Besides, VENUS is not limited by inherent domain knowledge and is adaptive to large-scale networks. Through experiments on multiple different public data sources, it demonstrates that VENUS is effective and effcient.

Exploring disease-associated human cells can help researchers identify pathogenesis and develop treatments. However, mining the relationships between diseases and human cells is difficult due to several challenges, including information extraction from unstructured knowledge sources. Biomedicine literature on diseases and cells contains a wealth of potentially useful information but is underutilized due to the complexity of unstructured information. A knowledge graph provides organized and structured relationships among entities and can be served as a solution. Nevertheless, few existing tools can be found for constructing a disease-aware human cell knowledge graph from biomedicine literature. To fill this gap, we propose a novel approach, called DEAL, to construct a disease-aware human cell knowledge graph from biomedicine literature. Technically, DEAL builds the knowledge graph through a series of steps including the processing of distant supervision data, the training of a distant supervision relation extraction model, the extraction of triples, and the knowledge graph construction. Experiments demonstrate that DEAL can effectively express the extracted entities related to diseases and cells and their relationships.

Drug-target interaction (DTI) prediction serves as the foundation of new drug findings and drug repositioning. For drugs/targets, the sequence data contains the biological structural information, while the heterogeneous network contains the biochemical functional information. These two types of information describe different aspects of drugs and targets. Due to the complexity of DTI machinery, it is necessary to learn the representation from multiple perspectives. We hereby try to design a way to leverage information from multi-source data to the maximum extent and find a strategy to fuse them. To address the above challenges, we propose a model, named MOVE (short for integrating multi-source information for predicting DTI via cross-view contrastive 1earning), for learning comprehensive representations of each drug and target from multi-source data. MOVE extracts information from the sequence view and the network view, then utilizes a fusion module with auxiliary contrastive learning to facilitate the fusion of representations. Experimental results on the benchmark dataset demonstrate that MOVE is effective in DTI prediction.

Retrieving similar products is a critical functionality required by many e-commerce websites as well as dynamic product advertising systems. Retargeting and Prospecting are two major forms of dynamic product advertising. Typically, after a user interacts with a product on an advertiser website (e.g., Macy's), when the user later visits a website (e.g., yahoo.com) supported by a dynamic product advertising system, the same product may be shown to the user as a Retargeting product ad, while some similar products may be shown to the user as Prospecting product ads on the web page. Similar products can enrich users' ad experience based on users' intent on the Prospecting product ads through which the users interacted. These product ads can also serve as substitutes when Retargeting ad candidates are out of stock. However, it is challenging to retrieve similar products among billions of products in a product catalog efficiently. Deep Siamese models allow efficient retrieval but do not put enough emphasize on key product attributes. To improve the quality of the similar products, we propose to first use a Siamese Transformer-based model to retrieve similar products and then refine them with the attribute "product name" that indicates the type of a product (e.g., running shoes, engagement ring, etc.) for post filtering. We propose a novel product name generation model that fine tunes a pre-trained Transformer-based language model with a sequence to sequence objective. To the best of our knowledge, this is the first work using a generative approach for identifying product attributes. We introduce two applications of the proposed approach for the dynamic product advertising system of Yahoo for Retargeting and Prospecting respectively. Offline evaluation and online A/B testing shows that the proposed approach retrieves high quality similar products, leading to an increase of ad clicks and ad revenue.

Predicting molecular properties has significant implications for the discovery and generation of drugs and further research in the domain of medicinal chemistry. Learning representations of molecules plays a central role in deep learning-driven property prediction. However, the diversity of molecular features (e.g., chemical system languages, structure notations) brings inconsistency in molecular representation. Moreover, the scarcity of labeled molecular data limits the accuracy of the molecular property prediction model. To address the above issues, we proposed a two-stage method, named MORN, for learning molecular representations for molecular property prediction from a multi-view perspective. In the first stage, textual-topological-spatial multi-views were proposed to learn the molecular representations, so as to capture both chemical system language and structure notation features simultaneously. In the second stage, an adaptive strategy was used to fuse molecular representations learned from multi-views to predict molecular properties. To alleviate the limitation of the scarcity of labeled molecular data, the label restriction was introduced in both multi-view representation learning and fusion stages. The performance of MORN was assessed by seven benchmark molecular datasets and one self-built molecular dataset. Experimental results demonstrated that MORN is effective in molecular property prediction.

Exploring efficient and high-accuracy computational drug repositioning methods has become a popular and attractive topic in drug development. This technology can systematically identify potential drug-disease interactions, which could greatly alleviate the pressures from the high cost and long period taken by traditional drug research and discovery. However, plenty of current computational drug repositioning approaches lack interpretability in predicting drug-disease associations, which will not be friendly to their subsequent in-depth research. To this end, we hereby propose a novel computational framework, called EDEN, for exploring explainable drug repositioning from the disease information network (DIN). EDEN is a graph neural network framework that learns the local semantics and global structure of the DIN, and models the drug-disease associations into the DIN by maximizing the mutual information of both and an end-to-end manner. In this way, the learned biomedical entity and link embeddings are enabled to retain the ability to drug repositioning with the semantical structure of external knowledge, thereby making interpretation possible. Meanwhile, we also propose a matching score based on the final embeddings to generate the predictive drug repositioning explanation. Empirical results on the real-world dataset show that EDEN outperforms other state-of-the-art baselines on most of the metrics. Further studies reveal the effectiveness of the explainability of our approach.

Analysis of gene similarity not only can provide information on the understanding of the biological roles and functions of a gene, but may also reveal the relationships among various genes. In this paper, we introduce a novel idea of mining similar aspects from a gene information network, i.e., for a given gene pair, we want to know in which aspects (meta paths) they are most similar from the perspective of the gene information network. We defined a similarity metric based on the set of meta paths connecting the query genes in the gene information network and used the rank of similarity of a gene pair in a meta path set to measure the similarity significance in that aspect. A minimal set of gene meta paths where the query gene pair ranks the highest is a similar aspect, and the similar aspect of a query gene pair is far from trivial. We proposed a novel method, SCENARIO, to investigate minimal similar aspects. Our empirical study on the gene information network, constructed from six public gene-related databases, verified that our proposed method is effective, efficient, and useful.

The emerging COVID-19 variants lead to a new wave of infections, spreading more rapidly with more severe illnesses. The adaptive immune system plays an essential role in the control and clearance of viral infection and influences clinical outcomes. However, the understanding of the adaptive immune responses to COVID-19 is not sufficient, which impedes the development progress of treatments and vaccines. To address this issue, we proposed a machine-learning-based method (termed as VDJ-Seg-Miner) to mine the underlying associations between the V(D)J gene segments of the T cell receptor in personalized immune repertoires and COVID-19 disease characteristics for immune system analysis. Our VDJ-Seg-Miner can interpretively reveal multiple associations between the V(D)J gene segments and COVID-19 disease characteristics and assign confidence scores to indicate its confidence in each revealed association. Furthermore, experimental results based on the real-world dataset suggested that the identified associations were highly consistent with those reported in previous work.

To identify similar diseases has significant implications for revealing the etiology and pathogenesis of diseases and further research in the domain of biomedicine. Currently, most methods for the measurement of disease similarity utilize either associations of ontological disease concepts or functional interactions between disease-related genes. These methods are heavily dependent on the ontology, which are not always available, and the selection of datasets. Moreover, many methods suffer from a drawback that they only use a single metric to evaluate disease similarity from an individual data source, which may result in biased conclusions without consideration of other aspects. In this study, we proposed a novel ontology-independent framework, namely RADAR, for learning representations for diseases to deduce their similarities from an integrative perspective. By leveraging the associations between diseases and disease-related biomedical entities, a disease similarity network was built under various metrics. Then, a multi-layer disease similarity network was constructed by integrating multiple disease similarity networks derived from multiple data sources, where the representation learning was derived to provide a comprehensive evaluation of disease similarities. The performance of RADAR was assessed by a benchmark disease set and 100 random disease sets. Experimental results demonstrated that RADAR can detect similar diseases effectively.

Non-coding RNAs are gaining prominence in biology and medicine, as they play major roles in cellular homeostasis and disease. A large number of computational methods have been recently developed for the prediction of the relationship between ncRNAs and diseases, which can alleviate the time-consuming and labor-intensive exploration among biological experiments. However, such methods have mainly focused on the association between the disease and certain types of ncRNAs such as miRNA or circRNA, thereby ignoring the impact of the interactions among ncRNAs on the diseases. We hereby propose a novel approach called DRAMA for discovering disease-related circRNA-miRNA-mRNA axes from the disease-RNA information network we constructed. Our method, using graph convolutional network, learns the characteristic representation of each biological entity by propagating and aggregating local neighbor information based on the global structure of the network. And then we design a favorable measurement to infer disease-related circRNA-miRNA-mRNA axes based on the learned embeddings. To evaluate the effectiveness of DRAMA, we conduct experiments on real-world datasets. Further analysis reveals that DRAMA outperforms other state-of-the-art baselines on most of the metrics.

To detect similar diseases is meaningful for revealing pathogenesis, and predicting therapeutic drugs. Previous methods measure disease similarity almost according to the semantic on biomedical ontology or the function of disease-causing molecules. However, such methods mostly describe diseases from single information, which may lead to a biased description of the relationships among diseases. In this paper, we propose a novel approach, called MISSION, for measuring the disease similarity based on multimodal-information. MISSION enhances similar disease detection based on disease information network from three aspects, including disease ontology, attribute, and literature, therefore providing a comprehensive evaluation for disease similarity. Through experiments on real-world datasets, we demonstrate that MISSION is effective, efficient, and potentially useful. Further analysis shows that MISSION has the ability to detect similar diseases with varying degrees of rich information.

Recently, a lot of research work has been proposed in different domains to detect outliers and analyze the outlierness of outliers for relational data. However, while sequence data is ubiquitous in real life, analyzing the outlierness for sequence data has not received enough attention. In this article, we study the problem of mining outlying sequence patterns in sequence data addressing the question: given a query sequence s in a sequence dataset D , the objective is to discover sequence patterns that will indicate the most unusualness (i.e., outlierness) of s compared against other sequences. Technically, we use the rank defined by the average probabilistic strength ( aps ) of a sequence pattern in a sequence to measure the outlierness of the sequence. Then a minimal sequence pattern where the query sequence is ranked the highest is defined as an outlying sequence pattern. To address the above problem, we present OSPMiner, a heuristic method that computes aps by incorporating several pruning techniques. Our empirical study using both real and synthetic data demonstrates that OSPMiner is effective and efficient.

In this paper we study the min-max cycle cover problem with neighborhoods, which is to find a given number of K cycles to collaboratively visit n Points of Interest (POIs) in a 2D space such that the length of the longest cycle among the K cycles is minimized. The problem arises from many applications, including employing mobile sinks to collect sensor data in wireless sensor networks (WSNs), dispatching charging vehicles to recharge sensors in rechargeable sensor networks, scheduling Unmanned Aerial Vehicles (UAVs) to monitor disaster areas, etc. For example, consider the application of employing multiple mobile sinks to collect sensor data in WSNs. If some mobile sink has a long data collection tour while the other mobile sinks have short tours, this incurs a long data collection latency of the sensors in the long tour. Existing studies assumed that one vehicle needs to move to the location of a POI to serve it. We however assume that the vehicle is able to serve the POI as long as the vehicle is within the neighborhood area of the POI. One such an example is that a mobile sink in a WSN can receive data from a sensor if it is within the transmission range of the sensor (e.g., within 50 meters). It can be seen that the ignorance of neighborhoods will incur a longer traveling length. On the other hand, most existing studies only took into account the vehicle traveling time but ignore the POI service time. Consequently, although the length of some vehicle tour is short, the total amount of time consumed by a vehicle in the tour is prohibitively long, due to many POIs in the tour. In this paper we first study the min-max cycle cover problem with neighborhoods, by incorporating both neighborhoods and POI service time into consideration. We then propose novel approximation algorithms for the problem, by exploring the combinatorial properties of the problem. We finally evaluate the proposed algorithms via experimental simulations. Experimental results show that the proposed algorithms are promising. Especially, the maximum tour times by the proposed algorithms are only about from 80% to 90% of that by existing algorithms.

In recent years, heatmap regression based models have shown their effectiveness in face alignment and pose estimation. However, Conventional Heatmap Regression (CHR) is not accurate nor stable when dealing with high-resolution facial videos, since it finds the maximum activated location in heatmaps which are generated from rounding coordinates, and thus leads to quantization errors when scaling back to the original high-resolution space. In this paper, we propose a Fractional Heatmap Regression (FHR) for high-resolution video-based face alignment. The proposed FHR can accurately estimate the fractional part according to the 2D Gaussian function by sampling three points in heatmaps. To further stabilize the landmarks among continuous video frames while maintaining the precise at the same time, we propose a novel stabilization loss that contains two terms to address time delay and non-smooth issues, respectively. Experiments on 300W, 300VW and Talking Face datasets clearly demonstrate that the proposed method is more accurate and stable than the state-ofthe-art models.

With the rapid development of anti-tumor biomaterials, biomedicine literature with respect to anti-tumor biomaterials has been leveraged for tumor treatment as it provides abundant and useful information. A large number of biomedicine literature contains unstructured data, making it difficult for researchers to obtain desired messages from it. Knowledge Graphs (KGs) provides structured relationships among entities and can be served as a solution. However, no existing tool can be found in constructing an anti-tumor biomaterial knowledge graph from biomedicine literature. To fill this gap, a novel approach, ATOM, was proposed to construct an anti-tumor biomaterial knowledge graph from biomedicine literature through a series of process including the recognition of anti-tumor entities, the simplification of sentences, the extraction of triples, and the predicate mapping. Experiments demonstrated that ATOM is able to effectively express the extracted anti-tumor entities and their relationships.

Gene similarity analysis not only provides information on understanding the biological roles and functions of a gene, but also reveals the relationships among different genes. In this paper, we identify the novel idea of mining similar aspects from gene information network, i.e., given a pair of genes, we want to know, in which aspects (meta paths) the two genes are mostly similar from the perspective of gene information network? We define a similarity metric based on the set of meta paths connecting the query genes in the gene information network, and use the rank of the similarity of a gene pair in a meta path set to measure the similarity significance in the aspect. A minimal set of meta paths where the query gene pair is ranked the best is a similar aspect. Computing the similar aspects of a query gene pair is far from trivial. In this paper, we propose a novel heuristic based-mining method, SCENARIO, to investigate minimal similar aspects. Our empirical study on the gene information network, constructed from seven public gene-related databases, verified that our proposed method is effective, efficient, and useful.

Unmanned aerial vehicles (UAVs) are emerging as promising devices to provide valuable information in rescue applications, which can be dispatched to take photographs for points of interests in disaster areas where humans are hard to approach. Most existing studies focused on the limited energy capacity issue of UAVs when they take photographs, which however ignored an important fact, that is, the photographs taken by the UAVs usually are highly redundant. In this paper we study a novel monitoring quality maximization problem to find a flying tour for an energy-constrained UAV, such that the amount of nonredundant information of the photographs taken by the UAV in its tour is maximized. Due to NP-hardness of the problem, we first propose an approximation algorithm with a quasi-polynomial time complexity. We then devise a fast yet scalable heuristic algorithm for the problem. We finally evaluate the performance of the proposed algorithms via both a real dataset and extensive simulations. Experimental results show that the proposed algorithms are very promising. Especially, the amounts of nonredundant information by the proposed approximation and heuristic algorithms are about 11% and 8% larger than that by the state-of-the-art, respectively. To the best of our knowledge, we are the first to consider the novel problem of collecting nonredundant information with an energy-constrained UAV.

Increasing the on-time rate of bus service can prompt the people’s willingness to travel by bus, which is an effective measure to mitigate the city traffic congestion. Performing queries on the bus arrival can be used to identify and analyze various kinds of non-on-time events that happened during the bus journey, which is helpful for detecting the factors of delaying events, and providing decision support for optimizing the bus schedules. We propose a data management model, called Bus-OLAP, for querying bus journey data, considering the characteristics of bus running and the scenarios of non-on-time analysis. While fulfilling typical requirements of bus journey data queries, Bus-OLAP not only provides a flexible way to manage the data and to implement multiple granularity data query and update, but it also supports distributed queries and computation. The experiments on real-world bus journey data verify that Bus-OLAP is effective and efficient.

Contrast patterns describe differences between two or more data sets or data classes; they have been proven to be useful for solving many kinds of problems, such as building accurate classifiers, defining clustering quality measures, and analyzing disease subtypes. This article investigates the mining of a new kind of contrast patterns, namelydiscriminating inter‐attribute functions(DIFs), which represent arithmetic‐expression‐based inter‐attribute relationships that distinguish classes of data. DIFs are an expressive and practical alternative of item‐based contrast patterns and can express discriminating relationships such as “weight/(height)2is more likely to be ≤25 in one class than in another class.” Besides introducing the DIF mining problem, this article makes theoretical and algorithmic contributions on the problem. We prove that DIF mining is MAX SNP‐hard. Regarding how to efficiently mine DIFs, we present a set of rules to prune the search space of arithmetic expressions by eliminating redundant ones (equivalent to some others). We give two algorithms: one for finding all DIFs satisfying given thresholds and another for finding certain optimal DIFs using genetic computation techniques. The former is useful when the number of attributes is small, whereas the latter is useful when that number is large; both use the redundant arithmetic‐expression pruning rules. A performance study shows that our techniques are effective and efficient for finding DIFs.

Relevant question retrieval and ranking is a typical task in community question answering (CQA). Existing methods mainly focus on long and syntactically structured queries. However, when an input query is short, the task becomes challenging, due to a lack information regarding user intent. In this paper, we mine different types of user intent from various sources for short queries. With these intent signals, we propose a new intent-based language model. The model takes advantage of both state-of-the-art relevance models and the extra intent information mined from multiple sources. We further employ a state-of-the-art learning-to-rank approach to estimate parameters in the model from training data. Experiments show that by leveraging user intent prediction, our model significantly outperforms the state-of-the-art relevance models in question search.

Function pointers have recently become an important attack vector for control-flow hijacking attacks. However, no protection mechanisms for function pointers have yet seen wide adoption. Methods proposed in the literature have high overheads, are not compatible with existing development process, or both. In this paper, we investigate several protection methods and propose a new method called FPGate (i.e., Function Pointer Gate). FPGate rewrites x86 binary executables and implements a novel method to overcome compatibility issues. All these protection methods are then evaluated and compared from the perspectives of performance and ease of deployment. Experiments show that FPGate achieves a good balance between performance, robustness and compatibility.

Control Flow Integrity (CFI) provides a strong protection against modern control-flow hijacking attacks. However, performance and compatibility issues limit its adoption. We propose a new practical and realistic protection method called CCFIR (Compact Control Flow Integrity and Randomization), which addresses the main barriers to CFI adoption. CCFIR collects all legal targets of indirect control-transfer instructions, puts them into a dedicated "Springboard section" in a random order, and then limits indirect transfers to flow only to them. Using the Springboard section for targets, CCFIR can validate a target more simply and faster than traditional CFI, and provide support for on-site target-randomization as well as better compatibility. Based on these approaches, CCFIR can stop control-flow hijacking attacks including ROP and return-into-libc. Results show that ROP gadgets are all eliminated. We observe that with the wide deployment of ASLR, Windows/x86 PE executables contain enough information in relocation tables which CCFIR can use to find all legal instructions and jump targets reliably, without source code or symbol information. We evaluate our prototype implementation on common web browsers and the SPEC CPU2000 suite: CCFIR protects large applications such as GCC and Firefox completely automatically, and has low performance overhead of about 3.6%/8.6% (average/max) using SPECint2000. Experiments on real-world exploits also show that CCFIR-hardened versions of IE6, Firefox 3.6 and other applications are protected effectively.

DCG (Dynamic Code Generation) technologies have found widely applications in the Web 2.0 era, Dion Blazakis recently presented a Flash JIT-Spraying attack against Adobe Flash Player that easily circumvented DEP and ASLR protection mechanisms built in modern operating systems. We have generalized and extended JIT Spraying into DCG Spraying. Based our analyses on this abstract model of DCG Spraying, we have found that all mainstream DCG implementations (Java/ JavaScript/ Flash/ .Net/ SilverLight) are vulnerable against DCG Spraying attack, and none of the existing ad hoc defenses such as compilation optimization, random NOP padding and constant splitting provides effective protection. Furthermore, we propose a new protection method, INSeRT, which combines randomization of intrinsic elements of machine instructions and randomly planted special trapping snippets. INSeRT practically renders the "sprayed code" ineffective, while alerts the host program of ongoing attacking attempts. We implemented a prototype of INSeRT on the V8 JavaScript engine, and the performance overhead is less than 5%, which should be acceptable in practical application.

Temporal information can be leveraged and incorporated to improve web search ranking. In this work, we propose a method to improve the ranking of search results by identifying the fundamental properties of temporal behavior of low-quality hosts and spam-prone queries in search logs and modeling those properties as quantifiable features. In particular, we introduce the concepts of host churn, a measure of changes in host visibility for user queries, and query volatility, a measure of semantic instability of query results, and propose the methods for construction of temporal profiles from search query logs that can be used for estimation of a set of features based on the introduced concepts. The utility of the proposed concepts has been experimentally demonstrated for two language-independent search tasks: the regression-based ranking of search results and a novel classification problem of detecting spam-prone queries introduced in this work.

A universal FPGA routing architecture is presented, which ensures that every module in the FPGA including CLBs and IOBs have a uniform interconnect architecture, and the load of interconnect lines is equally distributed. So, this architecture is highly repeatable and the signal delay is predictable and regular. Furthermore, the realization of the programmable interconnect point (PIP) and the buffer driver is also optimized to benefit the signal delay up to 5%.The test results of the example chip show the reasonableness of these ideas.

In this paper, we present a semi-supervised learning method for web page classification, leveraging click logs to augment training data by propagating class labels to unlabeled similar documents. Current state-of-the-art classifiers are supervised and require large amounts of manually labeled data. We hypothesize that unlabeled documents similar to our positive and negative labeled documents tend to be clicked through by the same user queries. Our proposed method leverages this hypothesis and augments our training set by modeling the similarity between documents in a click graph. We experiment with three different web page classifiers and show empirical evidence that our proposed approach outperforms state-of-the-art methods and reduces the amount of human effort to label training data.

We propose a novel cost-efficient approach to threshold selection for binary web-page classification problems with imbalanced class distributions. In many binary-classification tasks the distribution of classes is highly skewed. In such problems, using uniform random sampling in constructing sample sets for threshold setting requires large sample sizes in order to include a statistically sufficient number of examples of the minority class. On the other hand, manually labeling examples is expensive and budgetary considerations require that the size of sample sets be limited. These conflicting requirements make threshold selection a challenging problem. Our method of sample-set construction is a novel approach based on stratified sampling, in which manually labeled examples are expanded to reflect the true class distribution of the web-page population. Our experimental results show that using false positive rate as the criterion for threshold setting results in lower-variance threshold estimates than using other widely used accuracy measures such as F1 and precision.

In manufacturing grid, a lot of applications and resources in manufacturing enterprises are encapsulated as Web services and a cooperative work environment is provided. Applying Web service composition and workflow technologies, services are integrated into a composite service to represent a cooperative business process. Therefore the enterprise cooperation can be implemented by executing and monitoring the composite service. According to the characteristics of manufacturing, this paper presents a composite service modeling solution based on BPEL4WS and discusses how this solution support BPEL4WS during the whole modeling process. The solution includes dynamic service discovery and service selection to support a flexible binding with the partners' services. This paper also describes how to execute and monitor the composite service