Hate Video Detection (HVD) is crucial for online ecosystems. Existing methods assume identical distributions between training (source) and inference (target) data. However, hateful content often evolves into irregular and ambiguous forms to evade censorship, resulting in substantial semantic drift and rendering previously trained models ineffective. Test-Time Adaptation (TTA) offers a solution by adapting models during inference to narrow the cross-domain gap, while conventional TTA methods target mild distribution shifts and struggle with the severe semantic drift in HVD. To tackle these challenges, we propose SCANNER, the first TTA framework tailored for HVD. Motivated by the insight that, despite the evolving nature of hateful manifestations, their underlying cores remain largely invariant (i.e., targeting is still based on characteristics like gender, race, etc), we leverage these stable cores as a bridge to connect the source and target domains. Specifically, SCANNER initially reveals the stable cores from the ambiguous layout in evolving hateful content via a principled centroid-guided alignment mechanism. To alleviate the impact of outlier-like samples that are weakly correlated with centroids during the alignment process, SCANNER enhances the prior by incorporating a sample-level adaptive centroid alignment strategy, promoting more stable adaptation. Furthermore, to mitigate semantic collapse from overly uniform outputs within clusters, SCANNER introduces an intra-cluster diversity regularization that encourages the cluster-wise semantic richness. Experiments show that SCANNER outperforms all baselines, with an average gain of 4.69% in Macro-F1 over the best.

Internet congestion control remains a fundamental challenge, and recent learning-based congestion control algorithms (CCAs) have shown potential in optimizing network performance. However, their reliance on heuristically chosen input signals often leads to suboptimal behavior across diverse network conditions. In this paper, we identify the root cause as the lack of signal distinguishability-the ability of signals to reflect meaningful differences in network states. To address this, we propose Learn-to-Probe (LTP), a novel signal engineering paradigm that actively generates distinguishable network signals to improve the learning process. LTP (i) employs Bayesian filtering to accurately estimate network states from historical signals, and (ii) introduces an intrinsic reinforcement learning reward that encourages the flow to probe the network, inducing signal sequences that minimize the estimation uncertainty in (i). This probing behavior naturally enhances signal distinguishability, enabling the learning model to make more informed decisions. Extensive evaluations show that LTP consistently achieves high link utilization, low queuing delay, and stable convergence across diverse environments. Our results underscore the importance of signal distinguishability and offer a new direction for robust, adaptive congestion control.

Network traffic anomaly detection represents a critical cybersecurity task, yet widespread encryption makes this task increasingly challenging. In response, image-based methods that model traffic as visual patterns have emerged as the dominant approach. However, this work pioneers the identification of a pervasive "full-frequency"characteristic and an associated limitation termed "spectral mismatch"within this paradigm. Specifically, while encrypted traffic exhibits prominent high-frequency components, mainstream reconstruction methods demonstrate an inherent bias toward learning low-frequency information. This fundamental mismatch results in incomplete representations that consequently degrade anomaly detection performance. To address this challenge, we propose FreeUp, a novel frequency-decoupled framework designed explicitly for encrypted traffic analysis. FreeUp decomposes traffic data into distinct low- and high-frequency bands, processing them through separate, dedicated branches along with a customized training strategy that ensures stable and independent frequency-specific learning. Furthermore, recognizing that simple reconstruction error proves inadequate for evaluating dual-branch architectures, we introduce an uncertainty-inspired fusion scoring mechanism. This mechanism quantifies the reconstruction uncertainty of the frequency-specific branches and dynamically integrates their outputs, yielding a more comprehensive and reliable anomaly score. Extensive experiments across multiple benchmarks demonstrate that FreeUp consistently outperforms state-of-the-art baselines. The code is available at https://github.com/ikun0124/FreeUp.

Multimodal Large Language Model (MLLM) Personalization is a critical research problem that facilitates personalized dialogues with MLLMs targeting specific entities (known as personalized concepts). However, existing methods and benchmarks focus on the simple, context-agnostic visual identification and textual replacement of the personalized concept (e.g., "A yellow puppy" -> "Your puppy Mochi"), overlooking the ability to support long-context conversations. An ideal personalized MLLM assistant is capable of engaging in long-context dialogues with humans and continually improving its experience quality by learning from past dialogue histories. To bridge this gap, we propose LCMP, the first Long-Context MLLM Personalization evaluation benchmark. LCMP assesses the capability of MLLMs in perceiving variations of personalized concepts and generating contextually appropriate personalized responses that reflect these variations. As a strong baseline for LCMP, we introduce a novel training-free and state-aware framework TAME. TAME endows MLLMs with double memories to manage the temporal and persistent variations of each personalized concept in a differentiated manner. In addition, TAME incorporates a new training-free Retrieve-then-Align Augmented Generation (RA2G) paradigm. RA2G introduces an alignment step to extract the contextually fitted information from the multi-memory retrieved knowledge to the current questions, enabling better interactions for complex real-world user queries. Experiments on LCMP demonstrate that TAME achieves the best performance, showcasing remarkable and evolving interaction experiences in long-context scenarios.

Internet congestion control (CC) has long posed a challenging control problem in networking systems, with recent approaches increasingly incorporating deep reinforcement learning (DRL) to enhance adaptability and performance. Despite promising, DRL-based CC schemes often suffer from poor fairness, particularly when applied to network environments unseen during training. This paper introduces Jury, a novel DRL-based CC scheme designed to achieve fairness generalizability. At its heart, Jury decouples the fairness control from the principal DRL model with two design elements: i) By transforming network signals, it provides a universal view of network environments among competing flows, and ii) It adopts a post-processing phase to dynamically module the sending rate based on flow bandwidth occupancy estimation, ensuring large flows behave more conservatively and smaller flows more aggressively, thus achieving a fair and balanced bandwidth allocation. We have fully implemented Jury, and extensive evaluations demonstrate its robust convergence properties and high performance across a broad spectrum of both emulated and real-world network conditions.

With the rapid expansion of the internet, IP geolocation has become crucial for network security, content delivery, and compliance. However, existing methods struggle with dynamic, noisy networks, especially in handling topology changes, noisy data, and efficient node selection. To address these, we introduce PINGeo, a novel framework based on graph convolutional networks. PINGeo combines two innovations: (1) adaptive graph pruning using node importance metrics to retain critical nodes, optimizing both efficiency and accuracy, and (2) Gaussian noise perturbation to enhance robustness by simulating real-world fluctuations. By applying these methods, PINGeo improves geolocation accuracy and model robustness in noisy environments. Experimental results on public datasets show that PINGeo outperforms state-of-the-art methods in accuracy and robustness, offering a promising solution for robust IP geolocation.

Zero-shot composed image retrieval (ZS-CIR) is a challenging task that aims to retrieve images similar to a composed query of a reference image and a description, without relying on training on triplet datasets. Existing methods for this task often rely on predefined, fixed retrieval processes that combine the image and the modified text through hand-crafted templates, which suffer from two main issues: non-adaptive retrieval queries and user-unfriendly retrieval processes. To address these limitations, we propose a novel framework - Automatic Multi-Agent Collaboration for Zero-Shot Composed Image Retrieval (AutoCIR). AutoCIR consists of three training-free agents - a planner, a retriever, and a corrector - that work together to iteratively identify and rectify mismatches. The planner guides the retriever by generating a customized target caption for the composed query and further refines this caption to resolve any semantic discrepancies based on feedback. The corrector, equipped with a chain-of-thought reasoning mechanism, conducts an in-depth evaluation of the retrieved results and generates appropriate self-correction actions. Extensive experiments on three benchmarks demonstrate that AutoCIR consistently outperforms previous competitive methods for ZS-CIR.

Traditional multimodal learning approaches often assume that all modalities are available during both the training and inference phases. However, this assumption is often impractical in real-world scenarios due to challenges such as sensor failures, data corruption, or privacy concerns. While recent efforts focus on enhancing the robustness of pre-trained Multimodal Transformers (MTs) under missing modality conditions, mainstream work in this field often overlook reconstructing the missing modalities and rely on static, sample-agnostic prompt-tuning techniques, undermining their efficacy in severe modality missing scenarios. To address these limitations, we propose REDEEM, a novel REtrieval-guiDEd conditional gEnerative fraMework that largely alleviates the modality missing problems on pre-trained MTs. REDEEM consists of a new adaptive retrieval mechanism to identify relevant instances for both modality-complete and -incomplete samples. It then conditions on the remaining modalities and utilizes the retrieved data as experts to effectively recover the missing ones in modality-incomplete instances through a within-modal reconstruction manner. Finally, REDEEM generates sample-aware inter-modal prompts from the retrieved instances to guide MTs in tackling severe modality missing challenges. Comprehensive experiments on three diverse multimodal classification benchmarks demonstrate that REDEEM significantly outperforms competitive baselines.

Micro-video popularity prediction (MVPP) is a crucial research topic with important implications for social media marketing and stakeholders. Current works in MVPP utilized the pre-trained vision-language models (PVLs) to model the multimodal features for prediction, failing to capture the evolving popularity trend in micro-videos and leading to suboptimal results. To tackle this limitation, we propose EvoPro, an Evolution-aware Prompt-augmented framework that enhances MVPP. First, inspired by the powerful multimodal understanding and text generation skills of Large Multimodal Models (LMMs), an LMM-driven generative retriever is proposed to create contextually rich retrieval queries and perform precise video-to-video retrieval, forming dynamic micro-video support sets that effectively reflect evolving patterns. Building upon this, a graph-based prompter generates evolutionary prompts by capturing the relational structures within the support set. These prompts, representing the latest trend dynamics, serve as few-shot examples to guide PVLs. By integrating evolutionary prompts, the PVLs are empowered to model the evolving popularity trends more accurately, yielding stronger and more predictive representations. Extensive experiments conducted on three benchmarks demonstrate that EvoPro significantly outperforms competitive baselines.

Accurate IP geolocation is crucial for location-aware applications. While recent advances in router-centric IP graph methods have garnered attention, they face two persistent challenges: (1) the sparsity problem of IP graphs in rural areas and (2) the limited explainability of current IP geolocation systems. To tackle these issues, we present ExGeo, a novel and explainable graph-based approach for IP geolocation. Specifically, we introduce a target-centric IP graph, reducing sparsity and enhancing contextual information utilization. Additionally, we endow the model with explainability through a variational graph information bottleneck strategy. Experiments on three real-world datasets demonstrate significant accuracy and explainability improvements. Source code is released at https://github.com/ICDM-UESTC/ExGeo.

Trip recommendation has emerged as a highly sought-after service over the past decade. Although current studies significantly understand human intention consistency, they struggle with undesired repetitive outcomes that need resolution. We make two pivotal discoveries using statistical analyses and experimental designs: (1) The occurrence of repetitions is intricately linked to the models and decoding strategies. (2) During training and decoding, adding perturbations to logits can reduce repetition. Motivated by these observations, we introduce AR-Trip (Anti Repetition for Trip Recommendation), which incorporates a cycle-aware predictor comprising three mechanisms to avoid duplicate Points-of-Interest (POIs) and demonstrates their effectiveness in alleviating repetition. Experiments on four public datasets illustrate that AR-Trip successfully mitigates repetition issues while enhancing precision.

Point-of-Interest (POI) recommendation involves predicting users' next preferred POI and is becoming increasingly significant in location-based social networks. However, users are often reluctant to trust recommended results due to the lack of transparency in these systems. While recent work on explaining recommender systems has gained attention, prevailing methods only provide post-hoc explanations based on results or rudimentary explanations according to attention scores. Such limitations hinder reliability and applicability in risk-sensitive scenarios. Inspired by the information theory, we propose a self-explainable framework with an ante-hoc view called \M~for next POI recommendation aimed at overcoming these limitations. Specifically, we endow self-explainability to POI recommender systems through compact representation learning using a variational information bottleneck approach. The learned representation further improves accuracy by reducing redundancy behind massive spatial-temporal trajectories, which, in turn, boosts the recommendation performance. Experiments on three real-world datasets show significant improvements in both model explainability and recommendation performance.

Federated unlearning (FUL) is an emerging distributed machine learning paradigm which enables the removal or unlearning of specific training data effects from trained Federated Learning (FL) models. While current studies mostly focus on client-side FUL to address the “right to be forgotten”, and ignore the server's right to remove local models from the global model, particularly when clients are trained with low-quality data. In this paper, we introduce the Server-Initiated Federated Unlearning (SIFU) algorithm, devised to eliminate low-quality data from the global model. SIFU consists of two main components: (i) Identifying low-quality data: we develop a category-based method for quantifying low-quality data for each client and filter out clients containing such data. Datasets are then divided accordingly. (ii) Unlearning low-quality data: we employ gradient ascent training to counteract the adverse effects of low-quality data on local models. To minimize any bias introduced, we concurrently perform several batches of boosting training with good-quality data. SIFU could identify and promptly eliminate the impact of low-quality data on the FL global model while still preserving the benefits of good-quality data. Finally, extensive evaluations are conducted to verify the performance of SIFU with four different kinds of datasets and models. Results show that, compared to retraining from scratch, SIFU accelerates the speed of unlearning by 15× for small datasets (i.e., MNIST and FMNIST) and 20× for large datasets (i.e., CIFAR-10 and CelebA) without any degradation in accuracy, which also outperforms the state of the arts.

The rising popularity of online social network services has attracted a lot of research focusing on mining various user patterns. Among them, accurate IP geolocation is essential for a plethora of location-aware applications. However, despite extensive research efforts and significant advances, the "accurate and reliable'' desideratum is yet to be achieved at a higher quality level. This work presents a graph neural network (GNN)-based model, called TrustGeo, for trustworthy street-level IP geolocation. A distinct and important aspect of TrustGeo is the incorporation of sources of uncertainty in the learning process. The results of our extensive experimental evaluations on three real-world datasets demonstrate the superiority of our framework in significantly improving the accuracy and trustworthiness of street-level IP geolocation. Our code and datasets are available at https://github.com/ICDM-UESTC/TrustGeo.

IP geolocation refers to the process of determining the geographic locations of Internet Protocol (IP) addresses, which is important for mobile computing and spatial data management. Despite extensive research efforts, a client-independent geolocation service with high accuracy and reliability has not yet been developed. This paper presents a graph neural network (GNN) model, dubbed RIPGeo, for robust street-level IP geolocation. Three factors that affect data quality are identified, and the importance of considering data quality in algorithm development is emphasized. Two novel self-supervised perturbational training strategies are proposed to enhance the generalization and robustness of the model. A multi-task learning framework is introduced to solve the homogenized representation problem caused by perturbational training, demonstrating much more efficiency than prevailing solutions. Theoretical analysis and experimental results demonstrate the superiority of our framework in significantly improving the accuracy and stability of street-level IP geolocation.

Understanding users' behavior via IP addresses is essential towards numerous practical IP-based applications such as online content delivery, fraud prevention, and many others. Among which profiling IP address has been extensively studied, such as IP geolocation and anomaly detection. However, less is known about the scenario of an IP address, e.g., dedicated enterprise network or home broadband. In this work, we initiate the first attempt to address a large-scale IP scenario prediction problem. Specifically, we collect IP scenario data from four regions and propose a novel deep ensemble learning-based model to learn IP assignment rules and complex feature interactions. Extensive experiments support that our method can make accurate IP scenario identification and generalize from data in one region to another.

Pinpointing the geographic location of an IP address is important for a range of location-aware applications spanning from targeted advertising to fraud prevention. The majority of traditional measurement-based and recent learning-based methods either focus on the efficient employment of topology or utilize data mining to find clues of the target IP in publicly available sources. Motivated by the limitations in existing works, we propose a novel framework named GraphGeo, which provides a complete processing methodology for street-level IP geolocation with the application of graph neural networks. It incorporates IP hosts knowledge and kinds of neighborhood relationships into the graph to infer spatial topology for high-quality geolocation prediction. We explicitly consider and alleviate the negative impact of uncertainty caused by network jitter and congestion, which are pervasive in complicated network environments. Extensive evaluations across three large-scale real-world datasets demonstrate that GraphGeo significantly reduces the geolocation errors compared to the state-of-the-art methods. Moreover, the proposed framework has been deployed on the web platform as an online service for 6 months.

In theory, the variational auto-encoder (VAE) is not suitable for recommendation tasks, although it has been successfully utilized for collaborative filtering (CF) models. In this paper, we propose a Gaussian Copula-Vector Quantized Autoencoder (GC-VQAE) model that differs prior arts in two key ways: (1) Gaussian Copula helps to model the dependencies among latent variables which are used to construct a more complex distribution compared with the mean-field theory; and (2) by incorporating a vector quantisation method into encoders our model can learn discrete representations which are consistent with the observed data rather than directly sampling from the simple Gaussian distributions. Our approach is able to circumvent the "posterior collapse'' issue and break the prior constraint to improve the flexibility of latent vector encoding and learning ability. Empirically, GC-VQAE can significantly improve the recommendation performance compared to existing state-of-the-art methods.

The underlining task for fine-grained image recognition captures both the inter-class and intra-class discriminate features. Existing methods generally use auxiliary data to guide the network or a complex network comprising multiple sub-networks. They have two significant drawbacks: (1) Using auxiliary data like bounding boxes requires expert knowledge and expensive data annotation. (2) Using multiple sub-networks make network architecture complex and requires complicated training or multiple training steps. We propose an end-to-end Spatial Self-Attention Network (SSANet) comprising a spatial self-attention module (SSA) and a self-attention distillation (Self-AD) technique. The SSA encodes contextual information into local features, improving intra-class representation. Then, the Self-AD distills knowledge from the SSA to a primary feature map, obtaining inter-class representation. By accumulating classification losses from these two modules enables the network to learn both inter-class and intra-class features in one training step. The experiment findings demonstrate that SSANet is effective and achieves competitive performance.