Graph anomaly detection is emerging as a critical technology for addressing increasingly complex and dynamic risk environments. Although unsupervised graph anomaly detection has advanced under the graph representation learning, directly applying these paradigms remains fundamentally misaligned with anomaly detection objectives. In this work, we highlight two key insights: graph neural networks are often suboptimal as feature extractors due to neighborhood aggregation diluting anomaly signals, and reliance on local inconsistency mining is inadequate for comprehensive anomaly detection, as it often fails to identify anomalies hidden within camouflaged communities. Based on these insights, we propose multiscale inconsistency learning for graph anomaly detection (MI-GAD), a novel framework that integrates both local and global anomaly signals. Specifically, individual node representations are projected onto a common hypersphere to ensure uniformity. At the local scale, the graph structure is leveraged for affinity-aware modeling via group discrimination. At the global scale, we introduce node deviation, a metric that distinguishes anomalies by optimizing representation centers. This unified approach enables robust and comprehensive detection of diverse graph anomalies. Experiments on seven real datasets demonstrate that our method consistently outperforms state-of-the-art baselines in both effectiveness and scalability.

With the increasing scale and complexity of graph data, node attributes are also becoming richer and more complex, particularly in the form of informative text. Classic GNNs equipped with shallow attribute encoders are no longer sufficient to handle such data independently, making model collaboration across heterogeneous architectures an inevitable trend. Recently, the integration of Large Language Models (LLMs) and GNNs has attracted significant attention, yet the inherent disparity between these models remains a key challenge. Promising solutions have considered fine-tuning Small Language Models (SLMs) to bridge the gap between GNNs and frozen LLMs. However, this introduces another problem: these heterogeneous models bring complementary knowledge, but how to effectively integrate them and allow mutual refinement becomes a significant research gap. To address these challenges, we introduce COLA, a collaborative large–small model framework that enables seamless cooperation among semantic LLMs, task-specific fine-tuned SLMs, and structure-aware GNNs. COLA features a unique Consensus–Complement Coordination Mechanism (C3M), wherein its Mixture-of-Coordinators (MoC) architecturally aligns the LLM and SLM. Built upon this, a flexible graph-knowledge infusion strategy encourages the joint alignment and graph knowledge learning of textual representations. Extensive evaluations across nine diverse datasets show that COLA consistently achieves state-of-the-art performance, validating the effectiveness and generality of our collaborative paradigm.

Junhao Jia, Hanwen Zheng, Yueyi Wu, Huangwei Chen, Haishuai Wang, Jiajun Bu, Lei Wu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Large reasoning models (LRMs) achieve strong performance on complex tasks by generating intermediate reasoning before the final answer, yet they remain prone to reasoning hallucinations such as subtle arithmetic or constraintviolation errors.Prior hallucination detectors often rely on external verification or local tokenlevel signals, which are limited for LRMs and largely overlook whether the cross-phase information flow from reasoning to answering is structurally robust.We propose Routing Focus Score (RFS), a step-level indicator that measures how strongly cross-step attention routing aligns with semantic proximity derived from hidden-state cosine similarity.We further design RFS-Guard, a lightweight hallucination detection framework based on RFS.Empirically, we observe that higher reasoning-answer RFS is consistently associated with higher hallucination risk, suggesting a routing-collapse failure mode where models might prefer selfconfirmation loops and suppress the ability to audit their own generations.Experimental results across multiple domains and models demonstrate the superiority of RFS-Guard for detecting and localizing hallucinations in LRMs without requiring external tools or repeated sampling.

Long-term conversational agents rely on personal memory to maintain coherence and personalization, yet practical systems must operate under context budgets and cope with evolving or contradictory user information. We frame persona memory as a retrieval problem over a growing memory store, and propose REMAP, a reflection-guided memory editing approach for online alignment of persona facts that selectively writes and revises memory entries based on the current dialogue evidence and retrieved related items. The method aims to preserve salient facts while reducing redundancy and resolving apparent conflicts, enabling more efficient context utilization over extended interaction horizons. Experiments on multi-session dialogue datasets show consistent gains in persona-consistent retrieval and response continuity over commonly used memory strategies, while achieving more selective memory updates under comparable operational overhead.

This paper explores advancements in automated Question-Answer (QA) extraction using large language models (LLMs), addressing challenges in transforming unstructured text into high-quality, retrievable QA pairs. Traditional approaches, whether through segmented question and answer generation or end-to-end extraction, often struggle with efficiency, dataset limitations, and performance consistency. Leveraging recent progress in LLMs, we constructed a large-scale Chinese QA extraction dataset with 143,846 documents and evaluated multiple fine-tuned models on public and private datasets. Surprisingly, code-based English LLMs outperformed Chinese-specialized models on Chinese text with a lower hallucination rate. Building upon this finding, we enhanced the best-performing code-based model with an expanded Chinese vocabulary, creating Code Llama-M, which achieved better results. Integrating Code Llama-M into our internal assistant, Luo Ying, demonstrated notable user satisfaction gains, affirming its practical impact. Key contributions include: (i) creation of a robust Chinese QA extraction instruction dataset; (ii) evidence of cross-lingual efficacy of code-based LLMs for Chinese QA tasks, further enhanced through Code Llama-M's expanded Chinese vocabulary; and (iii) successful application of the fine-tuned LLM in a live assistant system, enhancing user experience.

Generative latent diffusion models (LDMs) have been extensively applied in various fields yet underperform in time-series prediction. Therefore, We propose the Re-Diffusion model, a latent diffusion approach that generates backbone residuals specifically tailored for time-series forecasting. The model comprises a variational autoencoder that compresses the residuals between the actual future values and the predictions from the backbone into latent space. It also includes a conditional diffusion generator to forecast the potential distribution of these residuals. Our findings reveal that this latent-space methodology particularly enhances existing backbone predictors, by effectively reducing prediction bias through an advanced estimation of complex error distributions. While previous diffusion-based models tend to struggle with long-term forecasting, Re-Diffusion integrates the strengths of diffusion methods, leading to improvements in long-term predictions. Our experimental results indicate that the Re-Diffusion model achieves a 10% promotion over state-of-art predictors, marking a significant advancement in the field of time-series forecasting.

With the proliferation of mobile sensing technology, substantial time series data have been generated and accumulated in various distributed domains, providing the basis for practical applications. Federated Learning (FL) has emerged as an essential framework for machine learning on decentralized data, especially with its potential for privacy-preserving. However, deployed with distributed and various edge devices, existing FL frameworks struggle to address the statistical heterogeneity of spatial feature distribution shifts from heterogeneous sensors. Due to the complex temporal dynamics of real-world time series data, temporal feature heterogeneity may result in suboptimal model adaptation performance. To this end, we propose a novelFederated learning approach withOrthogonal regularization forSpatial-Temporal heterogeneity (FedOST) on time series classification. It is featured in three aspects: (1) In the local training phase, we utilize an orthogonal projection to disentangle and align the shared and personalized features, as well as complementary information from different views of the time series data to formulate a robust multi-view training. (2) In the global aggregation phase, we adopt trainable global prototypes to improve feature space separation through orthogonal constraint, to serve as refined global knowledge in the local training. (3) In the testing phase, we leverage an uncertainty-aware test-time adaptation scheme to tackle the temporal feature shifts of unlabeled test data. We conduct extensive evaluations on real-world datasets, where FedOST outperforms existing state-of-the-art baselines with significant advantages.

Graph anomaly detection (GAD) refers to identifying abnormal graph nodes or edges that heavily deviate from normal observations. Existing approaches inevitably suffer from the influence of imbalanced data and privacy protection. This shortcoming poses challenges in optimizing node embeddings and detecting multitype anomalies simultaneously, resulting in decreased accuracy of existing GAD models. To address this shortcoming, we introduce a new federated learning model for graph anomaly detection (FedGAD). FedGAD enables collaborative unsupervised learning among decentralized data centers without requiring direct access to the distributed subgraphs. Specifically, FedGAD masks and reconstructs the neighborhood features to enhance the knowledge of node representations. Considering the data diversity across distributed clients, we also design a cross-clients' node representation module that enables nodes to reconstruct neighbors by leveraging information from other clients. Furthermore, we use a multiscale contrastive learning function, which includes both structure-level and contextual-level learning functions, to detect graph anomalies in the condition that subgraphs located at different clients show imbalanced data distributions. Experimental results on seven benchmark datasets demonstrate the superior performance of FedGAD compared with baseline methods, verifying its capability of improving GAD performance.

Personalized federated learning (PFL) on graphs is an emerging field focusing on the collaborative development of architectures across multiple clients, each with distinct graph data distributions while adhering to strict privacy standards. This area often requires extensive expert intervention in model design, which is a significant limitation. Recent advancements have aimed to automate the search for graph neural network architectures, incorporating large language models (LLMs) for their advanced reasoning and self-reflection capabilities. However, two technical challenges persist. First, although LLMs are effective in natural language processing, their ability to meet the complex demands of graph neural architecture search (GNAS) is still being explored. Second, while LLMs can guide the architecture search process, they do not directly solve the issue of client drift due to heterogeneous data distributions. To address these challenges, we introduce a novel method, Personalized Federated Graph Neural Architecture Search (PFGNAS). This approach employs a task-specific prompt to identify and integrate optimal GNN architectures continuously. To counteract client drift, PFGNAS utilizes a weight-sharing strategy of supernet, which optimizes the local architectures while ensuring client-specific personalization. Extensive evaluations show that PFGNAS significantly outperforms traditional PFL methods, highlighting the advantages of integrating LLMs into personalized federated learning environments.

Neural Representations for Videos (NeRV) has emerged as a promising implicit neural representation (INR) approach for video analysis, which represents videos as neural networks with frame indexes as inputs. However, NeRV-based methods are time-consuming when adapting to a large number of diverse videos, as each video requires a separate NeRV model to be trained from scratch. In addition, NeRV-based methods spatially require generating a high-dimension signal (i.e., an entire image) from the input of a low-dimension timestamp, and a video typically consists of tens of frames temporally that have a minor change between adjacent frames. To improve the efficiency of video representation, we propose Meta Neural Representations for Videos, named MetaNeRV, a novel framework for fast NeRV representation for unseen videos. MetaNeRV leverages a meta-learning framework to learn an optimal parameter initialization, which serves as a good starting point for adapting to new videos. To address the unique spatial and temporal characteristics of video modality, we further introduce spatial-temporal guidance to improve the representation capabilities of MetaNeRV. Specifically, the spatial guidance with a multi-resolution loss aims to capture the information from different resolution stages, and the temporal guidance with an effective progressive learning strategy could gradually refine the number of fitted frames during the meta-learning process. Extensive experiments conducted on multiple datasets demonstrate the superiority of MetaNeRV for video representations and video compression.

Graph Neural Networks (GNNs) have shown remarkable performance in modeling graph data. However, Labeling graph data typically relies on unreliable information, leading to noisy node labels. Existing approaches for GNNs under Label Noise (GLN) employ supervision signals beyond noisy labels for robust learning. While empirically effective, they tend to over-reliance on supervision signals built upon external assumptions, leading to restricted applicability. In this work, we shift the focus to exploring how to extract useful information and learn from the graph itself, thus achieving robust graph learning. From an information theory perspective, we theoretically and empirically demonstrate that the graph itself contains reliable information for graph learning under label noise. Based on these insights, we propose the Topological Feature Reconstruction (TFR) method. Specifically, TFR leverages the fact that the pattern of clean labels can more accurately reconstruct graph features through topology, while noisy labels cannot. TFR is a simple and theoretically guaranteed model for robust graph learning under label noise. We conduct extensive experiments across datasets with varying properties. The results demonstrate the robustness and broad applicability of our proposed TFR compared to state-of-the-art baselines. Codes are available at https://github.com/eaglelab-zju/TFR.

Recent advances in medical AI have shown a clear trend towards large models in healthcare. However, developing large models for multi-modal medical diagnosis remains challenging due to a lack of sufficient modal-complete medical data. Most existing multi-modal diagnostic models are relatively small and struggle with limited feature extraction capabilities. To bridge this gap, we propose AdaCoMed, an adaptive collaborative-learning framework that synergistically integrates the off-the-shelf medical single-modal large models with multi-modal small models. Our framework first employs a mixture-of-modality-experts (MoME) architecture to combine features extracted from multiple single-modal medical large models, and then introduces a novel adaptive co-learning mechanism to collaborate with a multi-modal small model. This co-learning mechanism, guided by an adaptive weighting strategy, dynamically balances the complementary strengths between the MoMEfused large model features and the cross-modal reasoning capabilities of the small model. Extensive experiments on two representative multi-modal medical datasets (MIMICIV-MM and MMIST ccRCC) across six modalities and four diagnostic tasks demonstrate consistent improvements over state-of-the-art baselines, making it a promising solution for real-world medical diagnosis applications. The code is available at https://github.com/Zoew420/AdaCoMed.

Detecting out-of-distribution (OOD) samples poses a significant safety challenge when deploying models in open-world scenarios. Advanced works assume that OOD and in-distributional (ID) samples exhibit a distribution discrepancy, showing an encouraging direction in estimating the uncertainty with embedding features or predicting outputs. In this work, we propose a data structure-aware approach to mitigate the sensitivity of distances to the “curse of dimensionality”, where high-dimensional features are mapped to the manifold of ID samples, leveraging the well-known manifold assumption. Specifically, we present a novel distance termed as tangent distance, which tackles the issue of generalizing the meaningfulness of distances on testing samples to detect OOD inputs. Extensive experiments show that the tangent distance performs competitively with other post hoc OOD detection baselines on common and large-scale benchmarks.

Recent advancements in test-time adaptation (TTA) offer promising solutions for mitigating performance degradation caused by distribution shifts. However, they may fall short in practical deployment of GNNs during test time, due to the significant reliance on impractical retraining or modifications to online GNN models. To address such challenges, in this work, we propose a novel method, i.e., Test-Time Graph REBirth, dubbed TT-GREB, to effectively generalize the well-trained GNN models to the test-time graphs under distribution shifts by directly manipulating the test graph data. Concretely, we develop an overall framework designed by two principles for obtaining newly reborn test graphs, corresponding to two sub-modules: (1) prototype extractor for re-extracting the environment-invariant features of the test-time graph; and (2) environment refiner for re-fining the environment-varying features to explore the potential shifts. Furthermore, we propose a dual test-time graph contrastive learning objective with an effective iterative optimization strategy to obtain optimal prototype components and environmental components of the test graph. Extensive experiments on real-world graphs under diverse test-time distribution shifts verify the effectiveness of our proposed method, showcasing its superior ability to manipulate test-time graphs for better GNN generalization ability.

Time series analysis is crucial across various fields like energy, environment, transportation, finance and health. Deep learning has significantly advanced this field, particularly, the Time Series Foundation Model (TSFM) excels in multiple domains due to extensive pre-training. In this work, we focus on TSFM’s challenges in medical practice: limited computing resources and medical data privacy. TSFM variants include fine-tuned models and those pre-trained for rapid deployment on diverse data. There may not be enough computing resources to train physiological signals locally in hospitals, and generalized TSFM is still inferior to task-specific methods on private, imbalanced local data. To address this, we propose PhysioPFM, a framework for efficiently personalizing TSFM. Our approach involves low-rank pre-training on public datasets, generator training by trained LoRA weights, and efficient weight generation via local data. Experimental results demonstrate that integrating generated models with TSFM enhances performance, and transferability, and reduces the need for additional sensitive data training.

Unsupervised Anomaly Detection (UAD) plays a crucial role in identifying abnormal patterns within data without labeled examples, holding significant practical implications across various domains. Although the individual contributions of representation learning and clustering to anomaly detection are well-established, their interdependencies remain under-explored due to the absence of a unified theoretical framework. Consequently, their collective potential to enhance anomaly detection performance remains largely untapped. To bridge this gap, in this paper, we propose a novel probabilistic mixture model for anomaly detection to establish a theoretical connection among representation learning, clustering, and anomaly detection. By maximizing a novel anomaly-aware data likelihood, representation learning and clustering can effectively reduce the adverse impact of anomalous data and collaboratively benefit anomaly detection. Meanwhile, a theoretically substantiated anomaly score is naturally derived from this framework. Lastly, drawing inspiration from gravitational analysis in physics, we have devised an improved anomaly score that more effectively harnesses the combined power of representation learning and clustering. Extensive experiments, involving 17 baseline methods across 30 diverse datasets, validate the effectiveness and generalization capability of the proposed method, surpassing state-of-the-art methods.

Healthcare data frequently contain a substantial proportion of missing values, necessitating effective time series imputation to support downstream disease diagnosis tasks. However, existing imputation methods focus on discrete data points and are unable to effectively model sparse data, resulting in particularly poor performance for imputing substantial missing values. In this paper, we propose a novel approach, ImputeINR, for time series imputation by employing implicit neural representations (INR) to learn continuous functions for time series. ImputeINR leverages the merits of INR in that the continuous functions are not coupled to sampling frequency and have infinite sampling frequency, allowing ImputeINR to generate fine-grained imputations even on extremely sparse observed values. Extensive experiments conducted on eight datasets with five ratios of masked values show the superior imputation performance of ImputeINR, especially for high missing ratios in time series data. We also validate that applying ImputeINR to impute missing values in healthcare data enhances the performance of downstream disease diagnosis tasks.

Integrating multi-omics datasets through data-driven analysis offers a comprehensive understanding of the complex biological processes underlying various diseases, particularly cancer. Graph Neural Networks (GNNs) have recently demonstrated remarkable ability to exploit relational structures in biological data, enabling advances in multi-omics integration for cancer subtype classification. Existing approaches often neglect the intricate coupling between heterogeneous omics, limiting their capacity to resolve subtle cancer subtype heterogeneity critical for precision oncology. To address these limitations, we propose a framework named Graph Transformer for Multi-omics Cancer Subtype Classification (GTMancer). This framework builds upon the GNN optimization problem and extends its application to complex multi-omics data. Specifically, our method leverages contrastive learning to embed multi-omics data into a unified semantic space. We unroll the multiplex graph optimization problem in that unified space and introduce dual sets of attention coefficients to capture structural graph priors both within and among multi-omics data. This approach enables global omics information to guide the refining of the representations of individual omics. Empirical experiments on seven real-world cancer datasets demonstrate that GTMancer outperforms existing state-of-the-art algorithms.

Graph learning has shown significant advantages in organizing and leveraging complex data, making it promising for numerous real-world applications with heterogeneous information, particularly multi-omics data analysis. Despite its potential in such scenarios, existing methods are still in their infancy, lacking architectural potential and struggling to handle such complex data. In this paper, we propose the Multiplex Mixture of Graph Learners (MMoG) framework. MMoG first conducts fine-grained processing of consensus and unique information, constructing consistent features and multiplex graph structures. Then, a macroscopically shared group of sub-GNNs with diverse orders and architectures synergistically learn representations, providing a foundation for strong interaction between different views. Inspired by the mixture of experts (MoE), each sample in different omics adaptively determines the neighborhood ranges and architectures for information aggregation, while blocking unsuitable sub-GNNs. MMoG treats the complex multi-omics analysis as a multi-view learning problem, and essentially decomposes it into multiple sub-problems, allowing each omics/view to solve intersecting yet unique sub-problem groups. Additionally, we introduce mutual information-driven orthogonal loss and balancing loss to avoid view collapse. Extensive experiments on multi-omics data across multiple cancer types highlight MMoG's superiority.

Graph Neural Networks (GNNs) have become powerful models for both node- and graph-level tasks. While node-level learning focuses on individual nodes and their local structures, graph-level learning encounters challenges in capturing the global properties of graphs. In this paper, we conduct a theoretical and experimental analysis of existing graph-level learning frameworks and find that these frameworks typically adopt a single-view perspective based solely on node degree, which limits their ability to capture comprehensive graph characteristics. To address these issues, we propose a multi-view approach that leverages different types of centrality measures to capture diverse aspects of graph structure. We design an attention-based mechanism to adaptively integrate these multiple views, and use it as a readout function to perform weighted summation of node embeddings, termed as Adaptive Centrality Readout (ACRead). ACRead demonstrates enhanced flexibility and effectiveness when integrated with various GNN architectures, outperforming state-of-the-art readout methods, including KerRead and Set Transformer. Additionally, this multi-view centrality approach can serve as a standalone graph-level learning framework without relying on GNNs, referred to as Adaptive Centrality-based Graph Learning (ACGL), which achieves competitive performance by effectively combining different centrality perspectives.

Time series anomaly detection aims to identify unusual patterns in data or deviations from systems' expected behavior. The reconstruction-based methods are the mainstream in this task, which learn point-wise representation via unsupervised learning. However, the unlabeled anomaly points in training data may cause these reconstruction-based methods to learn and reconstruct anomalous data, resulting in the challenge of capturing normal patterns. In this paper, we propose a time series anomaly detection method based on implicit neural representation (INR) reconstruction, named TSINR, to address this challenge. Due to the property of spectral bias, TSINR enables prioritizing low-frequency signals and exhibiting poorer performance on high-frequency abnormal data. Specifically, we adopt INR to parameterize time series data as a continuous function and employ a transformer-based architecture to predict the INR of given data. As a result, the proposed TSINR method achieves the advantage of capturing the temporal continuity and thus is more sensitive to discontinuous anomaly data. In addition, we further design a novel form of INR continuous function to learn inter- and intra-channel information, and leverage a pre-trained large language model to amplify the intense fluctuations in anomalies. Extensive experiments demonstrate that TSINR achieves superior overall performance on both univariate and multivariate time series anomaly detection benchmarks compared to other state-of-the-art reconstruction-based methods. Our codes are available here.

Dual-energy X-ray absorptiometry (DXA) enables accurate bone mineral density but requires specialized equipment and protocols. X-ray-based BMD screening offers opportunistic early detection, though prior methods struggle with X-ray intensity variations and demand large datasets. We introduce a cross-modal knowledge distillation BMD prediction framework (CMKD-BMD) fusing X-ray/CT data to enhance X-ray-only BMD prediction. Each single-modal student network employs a multi-scale visual extractor for hierarchical features, an unsupervised graph-based structural learner for anatomical relationships, and an adaptive fusion module to generate a unified representation. The teacher network integrates single-modal representations from students and transfers multimodal knowledge from the teacher to students. Our model outperforms existing methods on both the collected dataset (comprising 1620 X-rays and 280 CT cases) and the publicly available VerSe2019 dataset, demonstrating superior BMD estimation performance. Furthermore, we developed OrthoSim, an orthopedic surgical simulation platform with CMKD-BMD, which has shown promising clinical effectiveness in trial evaluations. Our code is available at https://github.com/KeyueShi/CMKD-BMD.

In recent years, graph-based multi-view learning has received widespread attention for its ability to utilize data dependencies to capture more comprehensive information from ubiquitous multi-view data. However, with the increase in data size and complexity, Euclidean space struggles to capture the hierarchical and exponentially expanding relationships of multi-view data in limited dimensions, leading to embedding distortion and insufficient cross-view alignment and interaction. To this end, we propose a hyperbolic multi-view heat diffusion method. Firstly, we utilize the negative curvature advantage of the hyperbolic space to construct a graph representation for each view separately, so that each view can still retain its hierarchical structure in relatively low dimensions. Then we construct a graph heat diffusion process on hyperbolic manifolds to ensure that each view is locally smoothed and globally aggregated while achieving semantic consistency through virtual views. We show that the method can be interpreted as a Riemannian gradient descent process for collaborative learning on hyperbolic manifolds, which not only effectively fuses multimodal information, but also significantly enhances the interaction and unified representation among different views. Experimental results show that the proposed framework achieves excellent performance in a variety of multi-view scenarios.

The past decade has witnessed rapid advancements in cross-modal retrieval, with significant progress made in accurately measuring the similarity between cross-modal pairs. However, the persistent hubness problem, a phenomenon where a small number of targets frequently appear as nearest neighbors to numerous queries, continues to hinder the precision of similarity measurements. Despite several proposed methods to reduce hubness, their underlying mechanisms remain poorly understood. To bridge this gap, we analyze the widely-adopted Inverted Softmax approach and demonstrate its effectiveness in balancing target probabilities during retrieval. Building on these insights, we propose a probability-balancing framework for more effective hubness reduction. We contend that balancing target probabilities alone is inadequate and, therefore, extend the framework to balance both query and target probabilities by introducing Sinkhorn Normalization (SN). Notably, we extend SN to scenarios where the true query distribution is unknown, showing that current methods, which rely solely on a query bank to estimate target hubness, produce suboptimal results due to a significant distributional gap between the query bank and targets. To mitigate this issue, we introduce Dual Bank Sinkhorn Normalization (DBSN), incorporating a corresponding target bank alongside the query bank to narrow this distributional gap. Our comprehensive evaluation across various cross-modal retrieval tasks, including image-text retrieval, video-text retrieval, and audio-text retrieval, demonstrates consistent performance improvements, validating the effectiveness of both SN and DBSN. All codes are publicly available at https://github.com/ppanzx/DBSN.

Osteoporosis has emerged as a significant global public health challenge, affecting more than 200 million people worldwide. Precise screening through bone mineral density (BMD) assessment is essential for timely intervention. Although dual-energy X-ray absorptiometry (DXA) serves as the gold standard for BMD evaluation, its widespread application is hampered by high costs and limited accessibility. In recent years, deep learning has demonstrated great potential in osteoporosis classification and predicting BMD from X-ray and CT scans. However, existing public datasets have notable limitations, particularly the absence of paired multimodal imaging datasets for cross-modal modeling and the lack of accurate BMD annotations required for osteoporosis-specific research and clinical model optimization. To address these issues, we introduce the Lumbar Multimodal Osteoporosis Screening dataset (LUMOS), the first multimodal dataset specifically designed for lumbar osteoporosis screening. LUMOS integrates clinical data from 803 patients, including 1,620 anteroposterior/lateral lumbar X-rays with BMD values and T-scores, comprehensive demographic information, and 280 lumbar CT scans. The advent of LUMOS is expected to propel forward research on automated osteoporosis classification, BMD prediction, and other related tasks. Its standardized and multimodal nature fills critical gaps in lumbar osteoporosis data, providing high-quality data support for the development and validation of medical AI algorithms in the early detection of osteoporosis. The dataset is available at https://keyueshi.github.io/LUMOS/.

The convergence of graph learning and multi-view learning has propelled the emergence of multi-view graph neural networks (MGNNs), offering strong capabilities to address complex real-world data characterized by heterogeneous yet interconnected information. While existing MGNNs exploit the potential of multi-view graphs, the inherent conflict persists between the two critical inductive biases of multi-view learning, consistency and complementarity. Consequently, the challenge of defining and resolving this tension in the new context of multi-view graphs remains largely underexplored. To bridge this gap, we propose Multi-view Collaborative Graph Experts (MvCGE), a novel framework grounded in the Mixture-of-Experts (MoE) paradigm. MvCGE establishes architectural consistency through shared parameters while preserving complementarity via layer-wise collaborative graph experts, which are dynamically activated by a graph-aware routing mechanism that adapts to the structural nuances of each view. This dual-level design is further reinforced by two novel components: a load equilibrium loss to prevent expert collapse and ensure balanced specialization, and a graph discrepancy loss based on distributional divergence to enhance inter-view complementarity. Extensive experiments on diverse datasets demonstrate MvCGE’s superiority.

Due to rapid advancements in deep learning, Transformer-based architectures have proven effective in speech emotion recognition (SER), largely due to their ability to model long-term dependencies more effectively than recurrent networks. The current Transformer architecture is not well-suited for SER because its large parameter number demands significant computational resources, making it less feasible in environments with limited resources. Furthermore, its application to SER is limited because human emotions, which are expressed in long segments of continuous speech, are inherently complex and ambiguous. Therefore, designing specialized Transformer models tailored for SER is essential. To address these challenges, we propose a novel knowledge distillation framework that combines meta-knowledge and curriculum-based distillation. Specifically, we fine-tune the teacher model to optimize it for the SER task. For the student model, we embed individual sequence time points into variable tokens, which are used to aggregate the global speech representation. Additionally, we combine supervised contrastive and cross-entropy loss to increase the inter-class distance between learnable features. Finally, we optimize the student model using both meta-knowledge and the curriculum-based distillation framework. Experimental results on two benchmark datasets, IEMOCAP and MELD, demonstrate that our method performs competitively with state-of-the-art approaches in SER.

Detecting out-of-distribution (OOD) samples poses a significant safety challenge when deploying models in open-world scenarios. Advanced works assume that OOD and in-distributional (ID) samples exhibit a distribution discrepancy, showing an encouraging direction in estimating the uncertainty with embedding features or predicting outputs. Besides incorporating auxiliary outlier as decision boundary, quantifying a “meaningful distance” in embedding space as uncertainty measurement is a promising strategy. However, these distances-based approaches overlook the data structure and heavily rely on the high-dimension features learned by deep neural networks, causing unreliable distances due to the “curse of dimensionality”. In this work, we propose a data structure-aware approach to mitigate the sensitivity of distances to the “curse of dimensionality”, where high-dimensional features are mapped to the manifold of ID samples, leveraging the well-known manifold assumption. Specifically, we present a novel distance termed astangent distance, which tackles the issue of generalizing the meaningfulness of distances on testing samples to detect OOD inputs. Inspired by manifold learning for adversarial examples, where adversarial region probability density is close to the orthogonal direction of the manifold, and both OOD and adversarial samples have common characteristic$-$imperceptible perturbations with shift distribution, we propose that OOD samples are relatively far away from the ID manifold, wheretangent distancedirectly computes the Euclidean distance between samples and the nearest submanifold space$-$instantiated as the linear approximation of local region on the manifold. We provide empirical and theoretical insights to demonstrate the effectiveness of OOD uncertainty measurements on the low-dimensional subspace. Extensive experiments show that thetangent distanceperforms competitively with other post hoc OOD detection baselines on common and large-scale benchmarks, and the theoretical analysis supports our claim that ID samples are likely to reside in high-density regions, explaining the effectiveness of internal connections among ID data.

With the rapid development of mobile network technology and wearable mobile devices, user-scenario interactions generate a large amount of user behavioral data in the form of multivariate event sequences. Due to data isolation, these multi-scenario events need to be jointly trained to achieve better prediction results. However, traditional federated learning methods face significant challenges when handling distributed event sequences. And the effectiveness of existing modeling approaches for event sequences in federated contexts has not been thoroughly explored. To this end, we propose Federated Multivariate Temporal Point Processes (FedMTPP), which enables learning from distributed event sequences within a novel federated learning framework and leverages efficient event modeling technology, MTPP, to forecast future events. Specifically, FedMTPP restores the temporal structure of the original event sequence by rearranging event embeddings and redesigns the autoregressive-based hidden representation computation in traditional MTPP, making it more suitable for federated prediction tasks. Additionally, FedMTPP incorporates advanced encryption techniques to effectively safeguard user privacy and security. Experimental results on both synthetic and real datasets demonstrate that FedMTPP substantially improves the performance of local models and achieves results comparable to state-of-the-art centralized MTPP methods.

Personalized federated learning (PFL) enhances the performance of customized client models through collaborative training without compromising data privacy and ownership. Some previous PFL methods rely on rich prior knowledge about the types of data heterogeneity (such as class imbalance or feature skew), which greatly limits their application ranges. In this paper, we study theUniversal Personalization in Federated Learning (UniPFL), the problem that has no prior knowledge about the types of data heterogeneity. In real-world PFL scenarios, UniPFL is potential because the data distributions of clients are usually heterogeneous and unknown to the server, where quantity imbalance, class imbalance, feature skew, or hybrid heterogeneity are possible contingencies. To address UniPFL, we proposeFedFD, a novel framework with local data augmentation and global concept fusion, which is based on the recent advances inthe foundation generative models(e.g., diffusion models, BLIP-2). On the client side, FedFD utilizes a diffusion model to assist local training by generating augmented data samples, and is then efficiently fine-tuned to be personalized. On the server side, we customize the aggregation strategies based on model similarities to learn both personalized models and diverse feature concepts. Extensive experiments show that FedFD reaches the state-of-the-art on (1) CIFAR-10 and CIFAR-100 for class imbalance; (2) DomainNet and Office-10 for feature skew, and (3) hybrid heterogeneity with both class and feature shifts.

Unsupervised graph anomaly detection aims at identifying rare patterns that deviate from the majority in a graph without the aid of labels, which is important for a variety of real-world applications. Recent advances have utilized graph neural networks (GNNs) to learn effective node representations by aggregating information from neighborhoods. This is motivated by the hypothesis that nodes in the graph tend to exhibit consistent behaviors with their neighborhoods. However, such consistency can be disrupted by graph anomalies in multiple ways. Most existing methods directly employ GNNs to learn representations, disregarding the negative impact of graph anomalies on GNNs, resulting in suboptimal node representations and anomaly detection performance. While a few recent approaches have redesigned GNNs for graph anomaly detection under semi-supervised label guidance, how to address the adverse effects of graph anomalies on GNNs in unsupervised scenarios and learn effective representations for anomaly detection are still underexplored. To bridge this gap, in this article, we propose a simple, yet effective framework for guarding GNNs for unsupervised graph anomaly detection (G3AD). Specifically, G3AD first introduces two auxiliary networks along with correlation constraints to guard the GNNs against inconsistent information encoding. Furthermore, G3AD introduces an adaptive caching (AC) module to guard the GNNs from directly reconstructing the observed graph data that contains anomalies. Extensive experiments demonstrate that our G3AD can outperform 20 state-of-the-art methods on both synthetic and real-world graph anomaly datasets, with flexible generalization ability in different GNN backbones.

Natural language inference (NLI) aims to infer the relationship between two texts: premise and hypothesis. However, many existing methods overlook the problem of overestimation of model performance due to superficial correlation biases in NLI datasets. We study this problem and find that most current models have taken NLI as one of the text-matching tasks, which ignores the asymmetry of the premise and hypothesis of NLI. Therefore, we propose a simple and effective augmentation method, Inversive-Reasoning Augmentation (IRA), to remove the superficial correlation bias. After training the different NLI models with our IRA-augmented data based on two widely-used NLI datasets, we observe more fair evaluation results of the performance and robustness of the various NLI models.

Dynamic graph data mining has gained popularity in recent years due to the rich information contained in dynamic graphs and their widespread use in the real world. Despite the advances in dynamic graph neural networks (DGNNs), the rich information and diverse downstream tasks have posed significant difficulties for the practical application of DGNNs in industrial scenarios. To this end, in this paper, we propose to address them by pre-training and present the Contrastive Pre-Training Method for Dynamic Graph Neural Networks (CPDG). CPDG tackles the challenges of pre-training for DGNNs, including generalization capability and long-short term modeling capability, through a flexible structural-temporal subgraph sampler along with structural-temporal contrastive pre-training schemes. Extensive experiments conducted on both large-scale research and industrial dynamic graph datasets show that CPDG outperforms existing methods in dynamic graph pre-training for various downstream tasks under three transfer settings.

Federated learning (FL) is undergoing significant traction due to its ability to perform privacy-preserving training on decentralized data. In this work, we focus on sensitive time series data collected by distributed sensors in real-world applications. However, time series data introduce the challenge of dual spatial-temporal feature skew due to their dynamic changes across domains and time, differing from computer vision. This key challenge includes inter-client spatial feature skew caused by heterogeneous sensor collection and intra-client temporal feature skew caused by dynamics in time series distribution. We follow the framework of Personalized Federated Learning (pFL) to handle dual feature drifts to enhance the capabilities of customized local models. Therefore, in this paper, we propose a method FedST to solve key challenges through orthogonal feature decoupling and regularization in both training and testing stages. During training, we collaborate time view and frequency view of time series data to enrich the mutual information and adopt orthogonal projection to disentangle and align the shared and personalized features between views, and between clients. During testing, we apply prototype-based predictions and model-based predictions to achieve model consistency based on shared features. Extensive experiments on multiple real-world classification datasets and multimodal time series datasets show our method consistently outperforms state-of-the-art baselines with clear advantages.

Recent neural information retrieval models using dense text representations generated by pre-trained models commonly face two issues. First, a pre-trained model (e.g., BERT) usually truncates a long document before giving its representation, which may cause the loss of some important semantic information. Second, although pre-training models like BERT have been widely used in generating sentence embeddings, a substantial body of literature has shown that the pre-training models often represent sentence embeddings in a homogeneous and narrow space, known as the problem of representation anisotropy, which hurts the quality of dense vector retrieval. In this paper, we split the query and the document in information retrieval into two sets of natural sentences and generate their sentence embeddings with BERT, the most popular pre-trained model. Before aggregating the sentence embeddings to get the entire embedding representations of the input query and document, to alleviate the usual representation degeneration problem of sentence embeddings from BERT, we sample the variational auto-encoder's latent space distribution to obtain isotropic sentence embeddings and utilize supervised contrastive learning to uniform the distribution of these sentence embeddings in the representation space. Our proposed model undergoes training optimization for both the query and the document in the abovementioned aspects. Our model performs well in evaluating three extensively researched neural information retrieval datasets.

Deep learning, neural architecture search, reinforcement learning, and embedded learning use scaled data, hyperspace, reward, and similarity to produce efficient converging neural networks. However, all these state-of-the-art frameworks require training to evaluate whether the examples are identical. Therefore, we propose our Probabilistic Asymmetric Convergence Network with Validation and Transform-Learning (PACNVT) framework that learns with fewer data, reduces the hyperspace with our validation technique and modified tangent activation function, reinforces the learning with our transform learning algorithms, and ascertains similarity independent of spatial and mask for transformational consistency. Furthermore, our framework generates neural networks that yield state-of-the-art accuracies on MNIST, OMNIGLOT, and CIFAR datasets. Moreover, representing regression as a similarity problem unveils previously unseen residual fluid intelligence patterns, yielding 99.92% accuracy with as little as 20 pairs.

Due to the objectivity of emotional expression in the central nervous system, EEG-based emotion recognition can effectively reflect humans' internal emotional states. In recent years, convolutional neural networks (CNNs) and recurrent neural networks (RNNs) have made significant strides in extracting local features and temporal dependencies from EEG signals. However, CNNs ignore spatial distribution information from EEG electrodes; moreover, RNNs may encounter issues such as exploding/vanishing gradients and high time consumption. To address these limitations, we propose an attention-based temporal graph representation network (ATGRNet) for EEG-based emotion recognition. Firstly, a hierarchical attention mechanism is introduced to integrate feature representations from both frequency bands and channels ordered by priority in EEG signals. Second, a graph convolutional neural network with top-k operation is utilized to capture internal relationships between EEG electrodes under different emotion patterns. Next, a residual-based graph readout mechanism is applied to accumulate the EEG feature node-level representations into graph-level representations. Finally, the obtained graph-level representations are fed into a temporal convolutional network (TCN) to extract the temporal dependencies between EEG frames. We evaluated our proposed ATGRNet on the SEED, DEAP and FACED datasets. The experimental findings show that the proposed ATGRNet surpasses the state-of-the-art graph-based mehtods for EEG-based emotion recognition.

Next destination recommendation is a crucial research area for understanding human travel behavior. However, existing studies often overlook the problem of underfitting, which arises due to the limited regularity in users' travel patterns. To tackle this issue, we leverage diverse co-occurrence patterns (CoPs) to discover potential user preferences. These patterns capture intersections with similar spatial and temporal characteristics in users' travels. However, traditional graph neural network (GNN)-based approaches struggle to effectively handle complex spatial-temporal CoPs. To overcome these challenges, we propose a novel framework called DHIN (Dynamic Heterogeneous Information Network). Firstly, to address the problem of underfitting, DHIN constructs intricate CoPs by leveraging abundant features and connection relationships. Additionally, to solve the needs of cold-start users, DHIN generates potential connections by capturing dynamic urban hotspots based on global users' travel trajectories. Moreover, to model dynamic heterogeneous information, DHIN utilizes a hierarchical attention mechanism and integrates a dynamic encoder. The mechanism integrates multi-level attention to learn informative embeddings from heterogeneous attributes and structures, while the dynamic encoder processes dynamic temporal information for updating node representations. Finally, extensive experiments conducted on real-world trajectory data demonstrate the effectiveness of the proposed DHIN model.

The goal of Speech Emotion Recognition (SER) is to enable computers to recognize the emotion category of a given utterance in the same way that humans do. The accuracy of SER is strongly dependent on the validity of the utterance-level representation obtained by the model. Nevertheless, the "dark knowledge" carried by non-target classes is always ignored by previous studies. In this paper, we propose a hierarchical network, called DKDFMH, which employs decoupled knowledge distillation in a deep convolutional neural network with a fused multi-head attention mechanism. Our approach applies logit distillation to obtain higher-level semantic features from different scales of attention sets and delve into the knowledge carried by non-target classes, thus guiding the model to focus more on the differences between sentiment features. To validate the effectiveness of our model, we conducted experiments on the Interactive Emotional Dyadic Motion Capture (IEMOCAP) dataset. We achieved competitive performance, with 79.1 % weighted accuracy (WA) and 77.1 % unweighted accuracy (UA). To the best of our knowledge, this is the first time since 2015 that logit distillation has been returned to state-of-the-art status.

Implicit neural representations (INRs) aim to learn a continuous function (i.e., a neural network) to represent an image, where the input and output of the function are pixel coordinates and RGB/Gray values, respectively. However, images tend to consist of many objects whose colors are not perfectly consistent, resulting in the challenge that image is actually a discontinuous piecewise function and cannot be well estimated by a continuous function. In this paper, we empirically investigate that if a neural network is enforced to fit a discontinuous piecewise function to reach a fixed small error, the time costs will increase exponentially with respect to the boundaries in the spatial domain of the target signal. We name this phenomenon the exponential-increase hypothesis. Under the exponential-increase hypothesis, learning INRs for images with many objects will converge very slowly. To address this issue, we first prove that partitioning a complex signal into several sub-regions and utilizing piecewise INRs to fit that signal can significantly speed up the convergence. Based on this fact, we introduce a simple partition mechanism to boost the performance of two INR methods for image reconstruction: one for learning INRs, and the other for learning-to-learn INRs. In both cases, we partition an image into different sub-regions and dedicate smaller networks for each part. In addition, we further propose two partition rules based on regular grids and semantic segmentation maps, respectively. Extensive experiments validate the effectiveness of the proposed partitioning methods in terms of learning INR for a single image (ordinary learning framework) and the learning-to-learn framework. Code is released here.

Surface subdivision is crucial in 3D mesh presentation and storage. Existing classical subdivision methods adopt manually-defined rules and fail to precisely restore a high-resolution mesh. Recently, a data-driven subdivision method delivers neural networks to learn complex non-linear subdivision schemes, but it suffers from time-consuming training data preparation and limited subdivision levels. In this paper, we propose a new data-driven surface subdivision framework, namely Implicit Neural Distance Optimization Neural Subdivision (INDONS), aiming to subdivide coarse meshes to restore high-resolution meshes. Our method subdivides coarse meshes once in each forward process while can be recursively applied to obtain multilevel subdivision results, which is the same as the classical subdivision mechanism. Moreover, we introduce a new loss function, named implicit neural distance, to provide precise supervised signals for training the subdivision network. The implicit neural distance is defined as the distance between a mesh and an implicit neural field. Extensive experiments demonstrate that our INDONS framework is flexible to be deployed, and INDONS outperforms classical subdivision methods and existing data-driven subdivision method in restoring high-resolution meshes.

Contrastive representation learning is crucial in medical time series analysis as it alleviates dependency on labor-intensive, domain-specific, and scarce expert annotations. However, existing contrastive learning methods primarily focus on one single data level, which fails to fully exploit the intricate nature of medical time series. To address this issue, we present COMET, an innovative hierarchical framework that leverages data consistencies at all inherent levels in medical time series. Our meticulously designed model systematically captures data consistency from four potential levels: observation, sample, trial, and patient levels. By developing contrastive loss at multiple levels, we can learn effective representations that preserve comprehensive data consistency, maximizing information utilization in a self-supervised manner. We conduct experiments in the challenging patient-independent setting. We compare COMET against six baselines using three diverse datasets, which include ECG signals for myocardial infarction and EEG signals for Alzheimer’s and Parkinson’s diseases. The results demonstrate that COMET consistently outperforms all baselines, particularly in setup with 10% and 1% labeled data fractions across all datasets. These results underscore the significant impact of our framework in advancing contrastive representation learning techniques for medical time series. The source code is available at https://github.com/DL4mHealth/COMET.

Wheelset fault detection with high accuracy is challenging due to poor image quality. Specifically, the wheelset images are collected dynamically outdoors and suffer from diffuse reflection and environmental interference. Thus, the images contain light stripe adhesions (light flairs) and local fractures to be inpainted. The existing inpainting models are inapplicable to restore grayscale wheelset images. They are also too heavy to be deployed in an embedded wheelset monitoring equipment. In this paper, we propose a lightweight high-precision inpainting model that consists of a recurrent similarity network with the ghost convolution (RSG-Net) to remove light flairs and repair local fractures. RSG-Net replaces standard Pconv (partial convolutional) layers with soft-coding ones that can improve the feature representational ability. To reduce the influence of the background region features on image restoration, an asymmetrical similarity measure is designed to calculate not only the angle difference between the target and the source feature vectors but also the activation of the source ones. The multi-scale structural similarity (MS-SSIM) loss term is introduced to precisely guide the structural information restoration, such as the stripe edges. Moreover, the ghost convolution is introduced in RSG-Net to realize the model compression that can retain the core features of wheelset images and remove the redundant features. We conduct three groups of experiments that demonstrate the accuracy superiority of the proposed RSG-Net over the baseline methods, and the number of parameters is reduced by about 50%.

Automated classification of respiratory sounds has become an active research area in recent years. While recent studies have utilised deep learning methods to aid with respiratory sound classification, the performance is heavily influenced by the datasets available for respiratory sound classification tasks, which tend to be smaller and imbalanced. In this paper, we propose to explore the effectiveness of a multi-branch Temporal Convolutional Network (TCN) architecture integrated with Squeeze-and-Excitation Network (SEnet), a system denoted herein as MBTCNSE, for respiratory sound classification. To the best of the authors’ knowledge, this is the first time that such a hybrid architecture has been employed for respiratory sounds classification. Experiments based on the ICBHI challenge respiratory sound dataset demonstrate the effectiveness of our method.

3D convolutional neural networks have revealed superior performance in processing volumetric data such as video and medical imaging. However, the competitive performance by leveraging 3D networks results in huge computational costs, which are far beyond that of 2D networks. In this paper, we propose a novel Hilbert curve-based cross-dimensionality distillation approach that facilitates the knowledge of 3D networks to improve the performance of 2D networks. The proposed Hilbert Distillation (HD) method preserves the structural information via the Hilbert curve, which maps high-dimensional (>=2) representations to one-dimensional continuous space-filling curves. Since the distilled 2D networks are supervised by the curves converted from dimensionally heterogeneous 3D features, the 2D networks are given an informative view in terms of learning structural information embedded in well-trained high-dimensional representations. We further propose a Variable-length Hilbert Distillation (VHD) method to dynamically shorten the walking stride of the Hilbert curve in activation feature areas and lengthen the stride in context feature areas, forcing the 2D networks to pay more attention to learning from activation features. The proposed algorithm outperforms the current state-of-the-art distillation techniques adapted to cross-dimensionality distillation on two classification tasks. Moreover, the distilled 2D networks by the proposed method achieve competitive performance with the original 3D networks, indicating the lightweight distilled 2D networks could potentially be the substitution of cumbersome 3D networks in the real-world scenario.

It is significant to accurately predict the epidemic trend of COVID-19 due to its detrimental impact on the global health and economy. Although machine learning based approaches have been applied to predict epidemic trend, standard models have shown low accuracy for long-term prediction due to a high level of uncertainty and lack of essential training data. This paper proposes an improved machine learning framework employing Generative Adversarial Network (GAN) and Long Short-Term Memory (LSTM) for adversarial training to forecast the potential threat of COVID-19 in countries where COVID-19 is rapidly spreading. It also investigates the most updated COVID-19 epidemiological data before October 18, 2020 and model the epidemic trend as time series that can be fed into the proposed model for data augmentation and trend prediction of the epidemic. The proposed model is trained to predict daily numbers of cumulative confirmed cases of COVID-19 in Italy, USA, China, Germany, UK, and across the world. Paper further analyzes and suggests which populations are at risk of contracting COVID-19.

Combining matrix factorization (MF) with network embedding (NE) has been a promising solution to social recommender systems. However, such a scheme suffers from the online predictive efficiency issue due to the ever-growing users and items. In this paper, we propose a novel hashing-based social recommendation model, called semi-discrete socially embedded matrix factorization (S2MF), which leverages the dual advantages of social information for recommendation effectiveness and hashing trick for online predictive efficiency. Experimental results demonstrate the advantages of S2MF over state-of-the-art discrete recommendation models and its real-valued competitors.

On-demand food delivery (OFD) platforms have greatly impacted the food service industry, where OFD recommendation systems play a central role in enhancing user experience and raising revenues. OFD recommendation, compared with existing online e-commerce recommendation systems, needs to put more emphasis on fulfillment time related variables, because the order fulfillment cycle time (OFCT) which refers to the time elapsed between a user placing a food order and receiving the food significantly influences a user's choice from the recommended items. In this paper, we investigate the OFCT related information and propose a Fulfillment-Time-Aware Personalized Ranking (FTAPR) method for recommendation. FTAPR mainly consists of three components. First, Transformers are used to estimate OFCT based on a large amount of user order sequences. Then, the predicted OFCT and other OFCT related features are fused and encoded by a deep & cross network to learn fulfillment time related feature representation. At the last step, the time bias representation from the deep & cross network is integrated into the ranking system to deliver final search results. Extensive offline and online experiments on real-world datasets collected from one of China's largest OFD platforms Ele.me show the superiority of our model, e.g., an online A/B testing shows that FTAPR brings 1.3% and 2.5% gains in CTR and CVR compared with baselines.

Malicious Web crawlers threaten information system due to heavily taking up bandwidth resources and stealing private user data. Ele.me, a prevalent on-demand food delivery platform in China, suffers from the negative impact of crawlers. The crawler detection systems face two major challenges: spatial patterns of the crawler behaviors and limited labeled data for training. In this paper, we present efficient solutions to tackle these challenges. Specifically, we propose a new Attributed Action Net (AANet for short) model to detect Location-Based Services~(LBS) crawlers and a three-stage learning framework to train the model. AANet consists of three different embedding modules, including the action token sequence, temporal-spatial attributes of users, and the context information of the raw data. We have deployed the model at Ele.me, and both offline experiments and online A/B tests show that the proposed method is superior to the state-of-the-art models for sequence data classification on the food delivery platform.

Live-streaming platforms have recently gained significant popularity by attracting an increasing number of young users and have become a very promising form of online shopping. Similar to the traditional online shopping platforms such as Taobao, live-streaming platforms also suffer from online malicious fraudulent behaviors where many transactions are not genuine. The existing anti-fraud models proposed to recognize fraudulent transactions on traditional online shopping platforms are inapplicable on live-streaming platforms. This is mainly because live-streaming platforms are characterized by a unique type of heterogeneous live-streaming networks where multiple heterogeneous types of nodes such as users, live-streamers, and products are connected with multiple different types of edges associated with edge features. In this paper, we propose a new approach based on a heterogeneous graph neural network for LIve-streaming Fraud dEtection (called LIFE). LIFE designs an innovative heterogeneous graph learning model that fully utilizes various heterogeneous information of shopping transactions, users, streamers, and items from a given live-streaming platform. Moreover, a label propagation algorithm is employed within our LIFE framework to handle the limited number of labeled fraudulent transactions for model training. Extensive experimental results on a large-scale Taobao live-streaming platform demonstrate that the proposed method is superior to the baseline models in terms of fraud detection effectiveness on live-streaming platforms. Furthermore, we conduct a case study to show that the proposed method is able to effectively detect fraud communities for live-streaming e-commerce platforms.

A crucial element of human–machine interaction, the automatic detection of emotional states from human speech has long been regarded as a challenging task for machine learning models. One vital challenge in speech emotion recognition (SER) is how to learn robust and discriminative representations from speech. Meanwhile, although machine learning methods have been widely applied in SER research, the inadequate amount of available annotated data has become a bottleneck that impedes the extended application of techniques (e.g., deep neural networks). To address this issue, we present a deep learning method that combines knowledge transfer and self-attention for SER tasks. Here, we apply the log-Mel spectrogram with deltas and delta-deltas as input. Moreover, given that emotions are time-dependent, we apply Temporal Convolutional Neural Networks (TCNs) to model the variations in emotions. We further introduce an attention transfer mechanism, which is based on a self-attention algorithm in order to learn long-term dependencies. The Self-Attention Transfer Network (SATN) in our proposed approach, takes advantage of attention autoencoders to learn attention from a source task, and then from speech recognition, followed by transferring this knowledge into SER. Evaluation built on the Interactive Emotional Dyadic Motion Capture (IEMOCAP) demonstrates the effectiveness of the novel model.

Logo classification has gained increasing attention for its various applications, such as copyright infringement detection, product recommendation and contextual advertising. Compared with other types of object images, the real-world logo images have larger variety in logo appearance and more complexity in their background. Therefore, recognizing the logo from images is challenging. To support efforts towards scalable logo classification task, we have curated a dataset, Logo-2K+, a new large-scale publicly available real-world logo dataset with 2,341 categories and 167,140 images. Compared with existing popular logo datasets, such as FlickrLogos-32 and LOGO-Net, Logo-2K+ has more comprehensive coverage of logo categories and larger quantity of logo images. Moreover, we propose a Discriminative Region Navigation and Augmentation Network (DRNA-Net), which is capable of discovering more informative logo regions and augmenting these image regions for logo classification. DRNA-Net consists of four sub-networks: the navigator sub-network first selected informative logo-relevant regions guided by the teacher sub-network, which can evaluate its confidence belonging to the ground-truth logo class. The data augmentation sub-network then augments the selected regions via both region cropping and region dropping. Finally, the scrutinizer sub-network fuses features from augmented regions and the whole image for logo classification. Comprehensive experiments on Logo-2K+ and other three existing benchmark datasets demonstrate the effectiveness of proposed method. Logo-2K+ and the proposed strong baseline DRNA-Net are expected to further the development of scalable logo image recognition, and the Logo-2K+ dataset can be found at https://github.com/msn199959/Logo-2k-plus-Dataset.

A growing area of mental health research is the search for speech-based objective markers for conditions such as depression. However, when combined with machine learning, this search can be challenging due to a limited amount of annotated training data. In this paper, we propose a novel crosstask approach which transfers attention mechanisms from speech recognition to aid depression severity measurement. This transfer is applied in a two-level hierarchical network which mirrors the natural hierarchical structure of speech. Experiments based on the Distress Analysis Interview Corpus - Wizard of Oz (DAIC-WOZ) dataset, as used in the 2017 Audio/Visual Emotion Challenge, demonstrate the effectiveness of our Hierarchical Attention Transfer Network. On the development set, the proposed approach achieves a root mean square error (RMSE) of 3.85, and a mean absolute error (MAE) of 2.99, on a Patient Health Questionnaire (PHQ)-8 scale [0], [24], while on the test set, it achieves an RMSE of 5.66 and an MAE of 4.28. To the best of our knowledge, these scores represent the best-known speech-only results to date on this corpus.

A fast-growing area of mental health research is the search for speech-based objective markers for conditions such as depression.One vital challenge in the development of speech-based depression severity assessment systems is the extraction of depression-relevant features from speech signals.In order to deliver more comprehensive feature representation, we herein explore the benefits of a hybrid network that encodes depressionrelated characteristics in speech for the task of depression severity assessment.The proposed network leverages self-attention networks (SAN) trained on low-level acoustic features and deep convolutional neural networks (DCNN) trained on 3D Log-Mel spectrograms.The feature representations learnt in the SAN and DCNN are concatenated and average pooling is exploited to aggregate complementary segment-level features.Finally, support vector regression is applied to predict a speaker's Beck Depression Inventory-II score.Experiments based on a subset of the Audio-Visual Depressive Language Corpus, as used in the 2013 and 2014 Audio/Visual Emotion Challenges, demonstrate the effectiveness of our proposed hybrid approach.

Graphs are commonly used to represent objects, such as images and text, for pattern classification. In a dynamic world, an object may continuously evolve over time, and so does the graph extracted from the underlying object. These changes in graph structure with respect to the temporal order present a new representation of the graph, in which an object corresponds to a set of time-variant graphs. In this paper, we formulate a novel time-variant graph classification task and propose a new graph feature, called a graph-shapelet pattern, for learning and classifying time-variant graphs. Graph-shapelet patterns are compact and discriminative graph transformation subsequences. A graph-shapelet pattern can be regarded as a graphical extension of a shapelet-a class of discriminative features designed for vector-based temporal data classification. To discover graph-shapelet patterns, we propose to convert a time-variant graph sequence into time-series data and use the discovered shapelets to find graph transformation subsequences as graph-shapelet patterns. By converting each graph-shapelet pattern into a unique tokenized graph transformation sequence, we can measure the similarity between two graph-shapelet patterns and therefore classify time-variant graphs. Experiments on both synthetic and real-world data demonstrate the superior performance of the proposed algorithms.

When users in online social networks make a decision, they are often affected by their neighbors. Social recommendation models utilize social information to reveal the impact of neighbors on user preferences, and this impact is often described by the linear superposition of neighbor preferences or by global trust propagation. Further exploration needs to be undertaken to determine whether the influence pattern of other users from online interaction behaviors is adequately described. In this paper, we introduce evolutionary opinion dynamics from the field of statistical physics into recommender systems, characterizing the impact of other users. We propose an opinion dynamic model by evolutionary game theory. To describe online user interactions, we define the strategies during an interaction between two users, and present the payoff for each strategy in terms of errors of estimated ratings. Therefore, user behaviors are associated with their preferences and ratings. In addition, we measure user influence according to their topological roles in the social network. We incorporate evolutionary opinion dynamics and user influence into the recommendation framework for the prediction of unknown ratings. Experiment results on two real-world datasets demonstrate that our method outperforms state-of the-art models in terms of accuracy, and it also performs well for cold-start users. Our method reduces the divergence of user preferences, in accordance with online opinion interactions. Furthermore, our method has approximate computational complexity with matrix factorization, and results in less computation than state-of-the-art models. Our method is quite general, and indicates that studies in social physics, statistics, and other research fields may be involved in recommendation to improve the performance.

Discrete speech emotion recognition (SER), the assignment of a single emotion label to an entire speech utterance, is typically performed as a sequence-to-label task.This approach, however, is limited, in that it can result in models that do not capture temporal changes in the speech signal, including those indicative of a particular emotion.One potential solution to overcome this limitation is to model SER as a sequence-to-sequence task instead.In this regard, we have developed an attention-based bidirectional long short-term memory (BLSTM) neural network in combination with a connectionist temporal classification (CTC) objective function (Attention-BLSTM-CTC) for SER.We also assessed the benefits of incorporating two contemporary attention mechanisms, namely component attention and quantum attention, into the CTC framework.To the best of the authors' knowledge, this is the first time that such a hybrid architecture has been employed for SER.We demonstrated the effectiveness of our approach on the Interactive Emotional Dyadic Motion Capture (IEMOCAP) and FAU-Aibo Emotion corpora.The experimental results demonstrate that our proposed model outperforms current state-of-the-art approaches.

This paper formulates the problem of learning discriminative features (i.e., segments) from networked time-series data, considering the linked information among time series. For example, social network users are considered to be social sensors that continuously generate social signals represented as a time series. The discriminative segments are often referred to as shapelets in a time series. Extracting shapelets for time-series analysis has been widely studied. However, existing works on shapelet selection assume that the time series are independent and identically distributed. This assumption restricts their applications to social networked time-series analysis since a user's actions can be correlated to his/her social affiliations. In this paper, we propose a novel network regularized least squares (NetRLS) feature selection model that combines typical time-series data and user network data for analysis. Experiments on real-world Twitter, Weibo, and DBLP networked time-series data demonstrate the performance of the proposed method. NetRLS performs better than the representative baselines on four evaluation criteria, namely classification accuracy, area under the curve (AUC), F1-score, and statistical significance analysis. NetRLS also has competitive running time as the baselines.

Automatic emotion recognition from speech, which is an important and challenging task in the field of affective computing, heavily relies on the effectiveness of the speech features for classification. Previous approaches to emotion recognition have mostly focused on the extraction of carefully hand-crafted features. How to model spatio-temporal dynamics for speech emotion recognition effectively is still under active investigation. In this paper, we propose a method to tackle the problem of emotional relevant feature extraction from speech by leveraging Attention-based Bidirectional Long Short-Term Memory Recurrent Neural Networks with fully convolutional networks in order to automatically learn the best spatio-temporal representations of speech signals. The learned high-level features are then fed into a deep neural network (DNN) to predict the final emotion. The experimental results on the Chinese Natural Audio-Visual Emotion Database (CHEAVD) and the Interactive Emotional Dyadic Motion Capture (IEMOCAP) corpora show that our method provides more accurate predictions compared with other existing emotion recognition algorithms.

With increased use of electronic medical records (EMRs), data mining on medical data has great potential to improve the quality of hospital treatment and increase the survival rate of patients. Early readmission prediction enables early intervention, which is essential to preventing serious or life-threatening events, and act as a substantial contributor to reduce healthcare costs. Existing works on predicting readmission often focus on certain vital signs and diseases by extracting statistical features. They also fail to consider skewness of class labels in medical data and different costs of misclassification errors. In this paper, we recur to the merits of convolutional neural networks (CNN) to automatically learn features from time series of vital sign, and categorical feature embedding to effectively encode feature vectors with heterogeneous clinical features, such as demographics, hospitalization history, vital signs, and laboratory tests. Then, both learnt features via CNN and statistical features via feature embedding are fed into a multilayer perceptron (MLP) for prediction. We use a cost-sensitive formulation to train MLP during prediction to tackle the imbalance and skewness challenge. We validate the proposed approach on two real medical datasets from Barnes-Jewish Hospital, and all data is taken from historical EMR databases and reflects the kinds of data that would realistically be available at the clinical prediction system in hospitals. We find that early prediction of readmission is possible and when compared with state-of-the-art existing methods used by hospitals, our methods perform significantly better. For example, using the general hospital wards data for 30-day readmission prediction, the area under the curve (AUC) for the proposed model was 0.70, significantly higher than all the baseline methods. Based on these results, a system is being deployed in hospital settings with the proposed forecasting algorithms to support treatment.

Multivariate time series (MTS) is useful for detecting abnormity cases in healthcare area. In this paper, we propose an ensemble boosting algorithm to classify abnormality surgery time series based on learning shapelet features. Specifically, we first learn shapelets by logistic regression from multivariate time series. Based on the learnt shapelets, we propose a MTS ensemble boosting approach when the time series arrives as stream fashion. Experimental results on a real-world medical dataset demonstrate the effectiveness of the proposed methods.

Graph ranking is a promising technique for image retrieval, but its effectiveness is limited by the so-called semantic gap. To mitigate this gap, clickthroughs, which are helpful to perceive the visual content of images, are adopted by graph ranking models recently. However, few existing models take both sparseness and noisiness of clickthroughs into account, which are important in refining the clickthrough-based search results. In this paper, we propose a novel solution to clickthrough refinement that consists of two key components. The first one is to prune the noisy clickthroughs by a neighbor voting strategy, and the second one is to enrich the incomplete clickthroughs by a Tri-CF (collaborative filtering) algorithm. Furthermore, we devise four solutions to affinity matrices fusion in order to leverage the clickthroughs and visual features within a unified graph ranking model. Our extensive experiments for clickthrough prediction and image retrieval validate the effectiveness of the proposed techniques in comparison to state-of-the-art approaches.

Network representation aims to represent the nodes in a network as continuous and compact vectors, and has attracted much attention in recent years due to its ability to capture complex structure relationships inside networks. However, existing network representation methods are commonly designed for homogeneous information networks where all the nodes (entities) of a network are of the same type, e.g., papers in a citation network. In this paper, we propose a universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space. The UNRA is built on our latest mutually updated neural language module, which simultaneously captures inter-relationship among homogeneous nodes and node-content correlation. Relationships between different types of nodes are also assembled and learned in a unified framework. Experiments validate that the UNRA achieves outstanding performance, compared to six other state-of-the-art algorithms, in node representation, node classification, and network visualization. In node classification, the UNRA achieves a 3% to 132% performance improvement in terms of accuracy.

In this paper we study a new problem of online discovering diffusion provenances in large networks. Existing work on network diffusion provenance identification focuses on offline learning where data collected from network detectors are static and a snapshot of the network is available before learning. However, an offline learning model does not meet the need for early warning, real-time awareness, or a real-time response to malicious information spreading in networks. To this end, we propose an online regression model for real-time diffusion provenance identification. Specifically, we first use offline collected network cascades to infer the edge transmission weights, and then use an online l 1 non-convex regression model as the identification model. The proposed methods are empirically evaluated on both synthetic and real-world networks. Experimental results demonstrate the effectiveness of the proposed model.

Graph classification is an important tool for analyzing data with structure dependency, where subgraphs are often used as features for learning. In reality, the dimension of the subgraphs crucially depends on the threshold setting of the frequency support parameter, and the number may become extremely large. As a result, subgraphs may be incrementally discovered to form a feature stream and require the underlying graph classifier to effectively discover representative subgraph features from the subgraph feature stream. In this paper, we propose a primal-dual incremental subgraph feature selection algorithm (ISF) based on a max-margin graph classifier. The ISF algorithm constructs a sequence of solutions that are both primal and dual feasible. Each primal-dual pair shrinks the dual gap and renders a better solution for the optimal subgraph feature set. To avoid bias of ISF algorithm on short-pattern subgraph features, we present a new incremental subgraph join feature selection algorithm (ISJF) by forcing graph classifiers to join short-pattern subgraphs and generate long-pattern subgraph features. We evaluate the performance of the proposed models on both synthetic networks and real-world social network data sets. Experimental results demonstrate the effectiveness of the proposed methods.

Most improvements for Naive Bayes (NB) have a common yet important flaw - these algorithms split the modeling of the classifier into two separate stages - the stage of preprocessing (e.g., feature selection and data expansion) and the stage of building the NB classifier. The first stage does not take the NB's objective function into consideration, so the performance of the classification cannot be guaranteed. Motivated by these facts and aiming to improve NB with accurate classification, we present a new learning algorithm called Evolutionary Local Instance Weighted Naive Bayes or ELWNB, to extend NB for classification. ELWNB combines local NB, instance weighted dataset extension and evolutionary algorithms seamlessly. Experiments on 20 UCI benchmark datasets demonstrate that ELWNB significantly outperforms NB and several other improved NB algorithms.

An information cascade occurs when a person observes the actions of others and then engages in the same acts. Cascades may break out if a large population of nodes in the network get affected. The outbreaks of cascades will often bring influential events, which leads to an open research problem: how to accurately predict the cascading outbreaks in social networks? Although there have been some existing works on cascading outbreak prediction, they ignore the structure information of cascades. In this paper, we propose to use subcascades as features for cascade outbreak prediction. We use frequent sequential pattern mining to extract subcascades and then propose a max-margin based classifier to select at most B features for prediction. The proposed model is empirically evaluated on both synthetic and real-world networks. Experimental results demonstrate the effectiveness of the proposed model.

Graph classification is an important tool for analysing structured and semi-structured data, where subgraphs are commonly used as the feature representation. However, the number and size of subgraph features crucially depend on the threshold parameters of frequent subgraph mining algorithms. Any improper setting of the parameters will generate many trivial short-pattern subgraph fragments which dominate the feature space, distort graph classifiers and bury interesting long-pattern subgraphs. In this paper, we propose a new Subgraph Join Feature Selection (SJFS) algorithm. The SJFS algorithm, by forcing graph classifiers to join short-pattern subgraph fragments, can defrag trivial subgraph features and deliver long-pattern interesting subgraphs. Experimental results on both synthetic and real-world social network graph data demonstrate the performance of the proposed method.

In this paper we study a new problem of online diffusion source detection in social networks. Existing work on diffusion source detection focuses on offline learning, which assumes data collected from network detectors are static and a snapshot of network is available before learning. However, an offline learning model does not meet the needs of early warning, real-time awareness, and real-time response of malicious information spreading in social networks. In this paper, we combine online learning and regression-based detection methods for real-time diffusion source detection. Specifically, we propose a new ℓ1non-convex regression model as the learning function, and an Online Stochastic Sub-gradient algorithm (OSS for short). The proposed model is empirically evaluated on both synthetic and real-world networks. Experimental results demonstrate the effectiveness of the proposed model.