Large visual-language models demonstrate exceptional transferability in biomedical image analysis, yet their robustness remains challenging in generalization scenarios such as few-shot learning. This limitation stems from an overly entangled prompt space, causing instability in the conditional distribution, coupled with saliency bias that drives attention to dominate key regions, thereby restricting the coverage of visual representations. Furthermore, cross-modal fusion often struggles to extract critical information signals from redundant alignments. To address these structural bottlenecks, we propose TriCLIP, a decomposition-based framework designed to reconstruct the entire information flow across prompt conditioning, visual evidence extraction, and cross-modal fusion. Prompt Distribution Regularization enhances robustness to linguistic variations by optimizing class-conditional random context perturbations for continuous prompts. Feedback-driven Counterfactual Masking constructs a feedback-driven reverse saliency view that suppresses dominant evidence to promote complementary cue learning and mitigate attention bias. Finally, Selective Regulation Fusion regulates fusion selectivity through paired reinforcement-suppression principles, amplifying information correspondence while suppressing redundant channels. Extensive experiments on diverse biomedical benchmarks indicate that the proposed TriCLIP consistently outperforms existing methods.
论文检索
输入标题、作者或关键词，从 990 篇学术成果中精准定位
Optimizing survival outcomes, such as patient survival or customer retention, is a critical objective in data-driven decision-making. Off-Policy Evaluation (OPE) provides a powerful framework for assessing such decision-making policies using logged data alone, without the need for costly or risky online experiments in high-stakes applications. However, typical estimators are not designed to handle right-censored survival outcomes, as they ignore unobserved survival times beyond the censoring time, leading to systematic underestimation of the true policy performance. To address this issue, we propose a novel framework for OPE and Off-Policy Learning (OPL) tailored for survival outcomes under censoring. Specifically, we introduce IPCW-IPS and IPCW-DR, which employ the Inverse Probability of Censoring Weighting technique to explicitly deal with censoring bias. We theoretically establish that our estimators are unbiased and that IPCW-DR achieves double robustness, ensuring consistency if either the propensity score or the outcome model is correct. Furthermore, we extend this framework to constrained OPL to optimize policy value under budget constraints. We demonstrate the effectiveness of our proposed methods through simulation studies and illustrate their practical impacts using public real-world data for both evaluation and learning tasks.
Transformer models are rapidly becoming a cornerstone of modern Internet of Things (IoT) applications, yet their computational and memory demands far exceed the capabilities of a single typical ultra-low-power IoT device. We present CATS, a framework for distributed transformer inference on ultra-low-power wireless devices, enabling multiple devices to collaboratively execute models far larger than what a single device can sustain. At its core, CATS is a communication-aware distributed transformer inference scheme co-designed across transformer partitioning, wireless communication and training. It employs SomeGather, a new pruned communication primitive that selectively broadcasts activation columns to reduce communication bandwidth and RAM usage without sacrificing model accuracy. Building on SomeGather, we design a partitioning method that exploits this primitive for efficient model parallelism. To cope with unreliable wireless communication, CATS employs message-dropout during training, which mimics packet losses and yields models that are robust to message loss during inference. In real-world experiments, we show that CATS brings distributed transformer inference to ultra-low-power wireless devices for the first time, with deployments on up to 16 devices that collaboratively execute transformer models up to 14 times larger than what a single device can run.
We study multi-task reinforcement learning (RL), a setting in which an agent learns a single, universal policy capable of generalising to arbitrary, possibly unseen tasks. We consider tasks specified as linear temporal logic (LTL) formulae, which are commonly used in formal methods to specify properties of systems, and have recently been successfully adopted in RL. In this setting, we present a novel task embedding technique leveraging a new generation of semantic LTL-to-automata translations, originally developed for temporal synthesis. The resulting semantically labelled automata contain rich, structured information in each state that allow us to (i) compute the automaton efficiently on-the-fly, (ii) extract expressive task embeddings used to condition the policy, and (iii) naturally support full LTL. Experimental results in a variety of domains demonstrate that our approach achieves state-of-the-art performance and is able to scale to complex specifications where existing methods fail.
Cross-Domain Graph Anomaly Detection supports the transfer of knowledge to unknown targets. Nevertheless, current approaches frequently struggle in the Cross-Broad-Domain paradigm which is characterized by two significant discrepancies: feature heterogeneity and structural disparity. Such diverse geometric topologies result in substantial manifold mismatch, as the traditional dependence on linear residuals in flat Euclidean space induces considerable geometric distortion when accommodating various topological patterns. We propose a novel framework termed Geometry-aware Riemannian Anomaly DEtector (GRADE) to tackle these difficulties. To address the semantic gap, we develop a domain-based contrastive learning technique combined with target-adaptive prompt tuning, which dynamically adjusts source representations to align with the target distribution while maintaining local contexts. To tackle the manifold mismatch, we present the Riemannian Residual Displacement (RRD), a curvature-adaptive metric that maps node representations onto multi-curvature Riemannian manifolds. By quantifying geodesic deviations in curved spaces, RRD corrects geometric distortions and identifies anomaly patterns that remain invariant to topological alterations. Additionally, a geometry-aware prototype alignment approach secures these residuals to data-independent priors for resilient inference. Comprehensive experiments on 11 benchmark datasets across 4 distinct domains demonstrate that GRADE significantly outperforms state-of-the-art baselines in both zero-shot and few-shot settings.
We study reinforcement learning (RL) in multi-dimensional continuous state and action spaces with one-sided feedback, where the agent receives partial observations of the state and obtains reward information for only a subset of the state-action space at each time step. This setting introduces substantial challenges in both learning efficiency and privacy preservation. To address these challenges, we propose POOL, a novel privacy-preserving RL algorithm. We conduct a comprehensive theoretical analysis of POOL, deriving a sample complexity bound of O~((1+E_rho) H^3 alpha^-2), which matches the known lower bounds for non-private RL. Here, E_rho denotes the privacy parameter, H is the time horizon, and alpha is optimality-gap parameter. Our findings show that it is possible to enforce strong privacy guarantees while maintaining high learning efficiency, marking a significant step toward practical, privacy-aware RL in multi-dimensional environments with one-sided feedback.
Compositional Zero-Shot Learning (CZSL) aims to combine known attributes and objects as primitives for recognizing previously unseen attribute-object pairs. Prior works either predict attributes and objects independently, missing their strong contextual dependency, or use unidirectional conditional modeling (e.g., object-guided attribute prediction), which is prone to error propagation. We propose PRPC, a Progressive Reasoning framework with Primitive Correction, which explicitly models the bidirectional dependency between attributes and objects via step-wise inference. PRPC performs mutual correction of primitives to suppress prediction errors in earlier steps. Specifically, we formulate CZSL as structured, Q&A-style Chain-of-Thought reasoning process and constrain the MLLM to follow predefined semantic steps to generate intermediate decisions. To further enhance the reliability and logical consistency of intermediate reasoning, we introduce reinforcement learning post-training with a GRPO-based objective, providing step-level rewards aligned with the progressive inference procedure. Extensive experiments on three CZSL benchmarks demonstrate that PRPC achieves state-of-the-art performance, validating the effectiveness of progressive reasoning and bidirectional correction for robust compositional generalization.
We present two new algorithms for the problem of lexicographic inference from conditional knowledge bases. These algorithms are based on SAT and MaxSAT encodings of the underlying problem of lexicographic comparisons of classical interpretations and require only a polynomial number of SAT solver calls. In our experimental evaluation we show that our new algorithms signficantly outperform the state of the art.
Reachability games are two-player games played on a graph, where the objective of REACH player is to reach the target set whereas the objective of SAFE player is to stay away from the target set. Reachability games have important applications in artificial intelligence and reactive synthesis, and many of these applications give rise to infinite-state reachability games. In this paper, we study turn-based reachability games on infinite-state graphs defined over valuations of a finite set of real variables. We consider the problem of determining the existence of and computing a winning strategy for REACH player. Our contributions are twofold. First, we propose ranking certificates for reachability games, a sound and complete proof rule for proving that REACH player has a winning strategy from the specified initial state. Second, we consider polynomial reachability games, where transitions and objectives are described by polynomial constraints over real variables, and propose a fully automated algorithm for computing a winning strategy for REACH player together with a formal correctness witness in the form of a ranking certificate. The algorithm is sound, semi-complete, and runs in sub-exponential time. Our experiments demonstrate the ability of our method to solve challenging examples from the literature that were out of the reach of existing methods. Specifically, for the classical Cinderella-Stepmother game, we are able to compute an optimal winning strategy for an arbitrary precision parameter for the first time.
Exploiting unlabeled data through semi-supervised learning (SSL) or leveraging pre-trained models via fine-tuning are two prevailing paradigms for addressing label-scarce scenarios. Recently, growing attention has been given to combining fine-tuning of pre-trained vision-language models (VLMs) with SSL, forming the emerging paradigm of semi-supervised fine-tuning. However, existing methods often suffer from model bias and hyperparameter sensitivity, due to reliance on prediction consistency or pre-defined confidence thresholds. To address these limitations, we propose a simple yet effective plug-and-play methodology named Bi-Consistency-Guided Self-Training (Bi-CoG), which assigns high-quality and low-bias pseudo-labels, by simultaneously exploiting inter-model and intra-model consistency, along with an error-aware dynamic pseudo-label assignment strategy. Both theoretical analysis and extensive experiments over 14 datasets demonstrate the effectiveness of Bi-CoG, which consistently and significantly improves the performance of existing methods.
Probabilistic time series forecasting seeks to quantify the uncertainty of future observations. While recent works introduce latent variables to alleviate the spurious dependencies caused by hidden confounders, thereby reducing overly wide confidence intervals, simply incorporating latent factors is not sufficient. When the underlying latent dynamics evolve at multiple temporal scales, existing methods may entangle temporally coarse and fine latent dynamics, which introduces spurious latent transitions and in turn amplifies predictive uncertainty. Therefore, disentangling temporally coarse and fine latent dynamics is essential for achieving sharper and more reliable probabilistic forecasting results. Building on this insight, we propose COFE (COarse And FinE latent dynamics disentanglement), a variational autoencoder–based framework that models the temporal distribution by disentangling and modeling the rapidly changing and slowly varying latent dynamics simultaneously. In particular, the proposed COFE harnesses the independence of estimated noises between adjacent latent states as well as a sparsity constraint on estimated noise to disentangle latent dynamics across different scales effectively. More specifically, we show that the multi-scale latent dynamics are disentangled with rigorous theoretical guarantees. Extensive experiments on 15 benchmark datasets, compared against 12 state-of-the-art baselines, demonstrate that COFE achieves superior uncertainty quantification, validating its effectiveness in a wide range of real-world scenarios. Code is available at https://github.com/polars8948/COFE
Time series forecasting (TSF) refers to a fundamental task of predicting future sequential data based on historical observations. One representative category of TSF methods is transformer-based approaches, which translate time series into token sequences (i.e., as raw texts) before applying well-established transformers. In this paper, we conducted extensive preliminary experiments to evaluate their stability against various noises, and we empirically observed an interesting phenomenon: applying noises with appropriate levels to training time series can promote the predictive accuracy of transformer-based TSF methods. We analyze this phenomenon from the perspective of token attention and find one basic reason is that applying noises with appropriate levels can lead to token attention weights wavy, which is more consistent to the basic characteristic of time series data. Motivated by this phenomenon and analysis, we apply perturbations and propose a sub-objective wrt, perturbations constraining token attention weights to be wavy. Accordingly, we propose a novel TSF method, namely Wave-Attention-aware TransformER (WATER). We empirically evaluate \baby across the commonly used benchmark datasets, and experimental results indicate that it consistently outperforms the existing TSF methods.
Multimodal recommender systems exploit visual and textual signals to alleviate data sparsity, but this also makes them more vulnerable to evasion-based promotion attacks. Existing defenses are largely limited to single-modal settings and mainly focus on poisoning-based threats, leaving evasion-based threats underexplored. In this work, we first identify a cross-modal gradient mismatch under the multi-user promotion setting, where visual and textual perturbations are optimized in inconsistent directions due to the dominance of distinct user groups. This phenomenon dilutes the attack effectiveness and leads robust training to underestimate worst-case risks. To address this issue, we propose Untargeted Adversarial Training with Multimodal Coordination (UAT-MC). UAT-MC tackles the challenge of unknown targeted items in evasion-based attacks (as opposed to poisoning-based attacks) by treating all items as potential targets, and introduces a gradient alignment mechanism to explicitly correct this mismatch. This design ensures synchronized perturbations across modalities, thereby maximizing adversarial strength for robust training. Extensive experiments demonstrate that UAT-MC significantly improves robustness against promotion attacks while maintaining acceptable recommendation performance under the defense–accuracy trade-off. Code is available at https://github.com/ gmXian/UAT-MC.
In many large language model (LLM) applications, generating structurally diverse candidate outputs is crucial for downstream decision-making and reasoning. However, conventional beam search allocates search budget at the token-prefix level, so multiple beam slots may be occupied by lexically different but structurally similar hypotheses. To address these limitations, we propose Grammar-State Aware Beam Search (GSA-Beam). Our method first organizes search at the constraint state level rather than individual prefix level, enabling explicit control over structural diversity. It further incorporates a dynamic state-level beam regulation mechanism that adaptively adjusts beam allocation based on the branching of constraint states, efficiently exploring the feasible structured solution space. Experiments on JSON-schema constrained generation show that GSA-Beam improves explicit state coverage and final structural diversity while preserving schema validity; across several LLM backbones, it reduces reference latency by 22.5% on average compared with standard constrained beam search. Code is available at https://github.com/Paradozile/GSABeam.
Recent deep learning-based ISP methods are primarily constrained by mimicking fixed camera pipelines, consequently struggling to achieve expert-level aesthetic quality. To this end, we propose AesISP, the first expert-level aesthetic ISP framework formulated as a reward flow model. Addressing the ill-posed nature of ISP, AesISP establishes a Privileged Prior Distillation paradigm. By leveraging expert-image-guided latent proxies, we decompose the intractable task into a tractable, proxy-driven learning process. Subsequently, we propose MeanFlow++, which reformulates the flow matching objective via target-prediction parameterization and a Progressive Temporal Curriculum. By evolving from learning instantaneous to long-range average velocities, this mechanism rectifies transport trajectories and uses deterministic mapping to impose explicit constraints, anchoring the generation trajectory to the expert manifold. Finally, we introduce a multi-dimensional reward-driven reinforcement learning approach.Leveraging Group Relative Policy Optimization (GRPO) to balance trade-offs across dimensions such as color and lighting, it steers the model to converge precisely on expert-level aesthetic standards. Experiments demonstrate that AesISP outperforms state-of-the-art methods in both quantitative metrics and aesthetic quality.
The diffusion auction is an emerging auction model which leverages social networks to recruit more buyers, thereby improving auction revenue and allocation efficiency. Previous studies on diffusion auctions have primarily focused on the single-parameter domain, where each buyer's valuation function is represented by a single private value. However, due to the complex interdependencies between allocations and network structures, few studies have examined the combinatorial auction domain, in which buyers' valuation functions exhibit multidimensionality. In this paper, we fully characterize implementable diffusion mechanisms within the combinatorial auction domain for the first time. Based on the characterization, we identify a class of allocation policies for each of which the optimal payment, namely the one maximizing the seller's revenue, can be derived in closed form. Furthermore, we apply the established theory to design a practical diffusion combinatorial auction, named the Exhausted-Reference Pricing Mechanism, which is incentive-compatible, individually rational, and weakly budget balanced.
In human-aware vision-and-language navigation (HA-VLN), agents need to follow instructions while navigating safely among people in crowded environments. Existing methods often struggle to effectively transform forward-looking risk prediction into decision-making basis, resulting in intelligent agents being prone to collisions in crowded scenarios and taking overly cautious actions, thereby reducing navigation efficiency. Moreover, it is difficult for high-level decisions based on social norms to be stably implemented in dynamic crowd scenarios, which can easily lead to behavioral jitter and planning deviation. In this work, we propose the AHead-VLN, a navigation framework that incorporates forward-looking risk information in the reasoning process. Specifically, we designed the path risk association (PRA) module, which compiles future risks into structured risk events aligned with each candidate path, making risk information decision-ready for the large language model (LLM) reasoning. To balance executability and decision stability, we have further designed the hierarchical LLM planner (HLP) with two specialized components. The ``Brain'' LLM makes route choices based on instructions and risks, while the ``Cerebellum'' LLM prepares executable action scripts for candidate routes and retrieves the appropriate script for execution. Experiments on the HA-VLN benchmark show that AHead-VLN improves success while enhancing safety compared to strong baselines.
Breaking the Trade-off: Orthogonal Semantic Decoupling for Generalizable and Fair Deepfake Detection
Deepfake detection faces dual challenges in real-world deployment: cross-domain generalization and demographic fairness. Existing approaches often struggle with a trade-off between these goals. Generalization-oriented detectors can over-rely on demographic shortcuts, while fairness constraints tend to steer optimization away from the most discriminative decision boundary. To address this, we propose Orthogonal Semantic Decoupling (OSD), a framework that decouples demographic semantics from forgery cues. Specifically, we perform Singular Value Decomposition on the pretrained weights of a vision-language model, freezing the principal semantic subspace while learning parameter-efficient low-rank experts in the residual subspace. The experts comprise (1) Demographic Semantic Experts, a set of experts specialized via hard sampling and routed based on the similarities between image embeddings and text embeddings of predefined descriptions; and (2) a Universal Forgery Expert, which captures forgery features transferable across domains and demographics. Extensive experiments across multiple benchmarks demonstrate that our approach outperforms state-of-the-art methods in both generalization and fairness, breaking the trade-off. The code is available at https://github.com/sonder-lin/osd-deepfake-detection.
With the explosive growth of text generated by large language models (LLMs) on the internet, technologies for detecting LLM-generated text are increasingly important. However, existing technologies often require frequent access to LLMs during detection, significantly restricting their application in scenarios where only the text is accessible. To tackle this challenge, we identify that human-written texts exhibit higher entropy than LLM-generated texts. Motivated by this, we propose a novel method named IED, which leverages the Information Entropy for generative text Detection. Specifically, IED first transforms the given text into a vector of which each dimension represents the information entropy of each word, and then adopts a classifier to conduct the detection. Further, we propose IEGD which adopts the information gain to construct the vector by enhancing the differentiation between human-written and LLM-generated texts. To evaluate the effective of the proposed method, we construct a new large-scale dataset comprised of generated texts using various LLMs (LLaMA-7B, GPT-NeoX, etc.). Extensive experiments show that the proposed methods achieve improvement over state-of-the-art methods by up to 10.81%.
Continual knowledge graph embedding (CKGE) has gained popularity for managing dynamic knowledge graphs. Unlike general graph continual-learning approaches, CKGE focuses on retaining triple-level knowledge, thereby overcoming the inability of static models to accommodate continuously arriving facts. However, as new fact triples come in, real-world knowledge graphs often experience structural distribution shifts: the underlying topology and the way facts interconnect evolve over time. This presents a challenge that the structural distribution of historical data may diverge from that of emerging data, creating a distributional mismatch that complicates the adaptation to new trends. To address this, we propose Pattern-Anchor Alignment (PAAL), which introduces relation-level structural anchors to explicitly model these structural shifts. By quantifying the alignment between the structural distribution of emerging triples and historical anchors, PAAL implements an alignment-modulated optimization strategy. It utilizes a calibrated gating mechanism to precisely regulate the intensity of historical constraints based on structural stability. Experimental results indicate the effectiveness of PAAL, showing average H@1 improvements of 11.77% and 9.29% on pattern-shift and standard CKGE benchmarks, respectively. Our code is available at https://github.com/cangjie553/PAAL.