Physics-Constrained Cross-Resolution Enhancement Network for Optics-Guided Thermal UAV Image Super-Resolution

Optics-guided thermal UAV image super-resolution has attracted significant research interest due to its potential in all-weather monitoring applications. However, existing methods typically compress optical features to match thermal feature dimensions for cross-modal alignment and fusion, which not only causes the loss of high-frequency information that is beneficial for thermal super-resolution, but also introduces physically inconsistent artifacts such as texture distortions and edge blurring by overlooking differences in the imaging physics between modalities. To address these challenges, we propose PCNet to achieve cross-resolution mutual enhancement between optical and thermal modalities, while physically constraining the optical guidance process via thermal conduction to enable robust thermal UAV image super-resolution. In particular, we design a Cross-Resolution Mutual Enhancement Module (CRME) to jointly optimize thermal image super-resolution and optical-to-thermal modality conversion, facilitating effective bidirectional feature interaction across resolutions while preserving high-frequency optical priors. Moreover, we propose a Physics-Driven Thermal Conduction Module (PDTM) that incorporates two-dimensional heat conduction into optical guidance, modeling spatially-varying heat conduction properties to prevent inconsistent artifacts. In addition, we introduce a temperature consistency loss that enforces regional distribution consistency and boundary gradient smoothness to ensure generated thermal images align with real-world thermal radiation principles. Extensive experiments on VGTSR2.0 and DroneVehicle datasets demonstrate that PCNet significantly outperforms state-of-the-art methods on both reconstruction quality and downstream tasks including semantic segmentation and object detection.

RobustMask: Certified Robustness against Adversarial Neural Ranking Attack via Randomized Masking

Neural ranking models have achieved remarkable progress and are now widely deployed in real-world applications such as Retrieval-Augmented Generation (RAG). However, like other neural architectures, they remain vulnerable to adversarial manipulations: subtle character-, word-, or phrase-level perturbations can poison retrieval results and artificially promote targeted candidates, undermining the integrity of search engines and downstream systems. Existing defenses either rely on heuristics with poor generalization or on certified methods that assume overly strong adversarial knowledge, limiting their practical use. To address these challenges, we propose RobustMask, a novel defense that combines the context-prediction capability of pretrained language models with a randomized masking-based smoothing mechanism. Our approach strengthens neural ranking models against adversarial perturbations at the character, word, and phrase levels. Leveraging both the pairwise comparison ability of ranking models and probabilistic statistical analysis, we provide a theoretical proof of RobustMask's certified top-K robustness. Extensive experiments further demonstrate that RobustMask successfully certifies over 20% of candidate documents within the top-10 ranking positions against adversarial perturbations affecting up to 30% of their content. These results highlight the effectiveness of RobustMask in enhancing the adversarial robustness of neural ranking models, marking a significant step toward providing stronger security guarantees for real-world retrieval systems.

PEAR: Phase Entropy Aware Reward for Efficient Reasoning

Large Reasoning Models (LRMs) have achieved impressive performance on complex reasoning tasks by generating detailed chain-of-thought (CoT) explanations. However, these responses are often excessively long, containing redundant reasoning steps that inflate inference cost and reduce usability. Controlling the length of generated reasoning without sacrificing accuracy remains an open challenge. Through a systematic empirical analysis, we reveal a consistent positive correlation between model entropy and response length at different reasoning stages across diverse LRMs: the thinking phase exhibits higher entropy, reflecting exploratory behavior of longer responses, while the final answer phase shows lower entropy, indicating a more deterministic solution.This observation suggests that entropy at different reasoning stages can serve as a control knob for balancing conciseness and performance. Based on this insight, this paper introduces Phase Entropy Aware Reward (PEAR), a reward mechanism that incorporating phase-dependent entropy into the reward design. Instead of treating all tokens uniformly, PEAR penalize excessive entropy during the thinking phase and allowing moderate exploration at the final answer phase, which encourages models to generate concise reasoning traces that retain sufficient flexibility to solve the task correctly. This enables adaptive control of response length without relying on explicit length targets or rigid truncation rules. Extensive experiments across four benchmarks demonstrate that PEAR consistently reduces response length while sustaining competitive accuracy across model scales. In addition, PEAR demonstrates strong out-of-distribution (OOD) robustness beyond the training distribution. Our code is available at: https://github.com/iNLP-Lab/PEAR.

Language Models Can Learn from Verbal Feedback Without Scalar Rewards

LLMs are often trained with RL from human or AI feedback, yet such methods typically compress nuanced feedback into scalar rewards, discarding much of their richness and inducing scale imbalance. We propose treating verbal feedback as a conditioning signal. Inspired by language priors in text-to-image generation, which enable novel outputs from unseen prompts, we introduce the feedback-conditional policy (FCP). FCP learns directly from response-feedback pairs, approximating the feedback-conditional posterior through maximum likelihood training on offline data. We further develop an online bootstrapping stage where the policy generates under positive conditions and receives fresh feedback to refine itself. This reframes feedback-driven learning as conditional generation rather than reward optimization, offering a more expressive way for LLMs to directly learn from verbal feedback. Our code is available at https://github.com/sail-sg/feedback-conditional-policy.

P/D-Device: Disaggregated Large Language Model between Cloud and Devices

Serving disaggregated large language models has been widely adopted in industrial practice for enhanced performance. However, too many tokens generated in decoding phase, i.e., occupying the resources for a long time, essentially hamper the cloud from achieving a higher throughput. Meanwhile, due to limited on-device resources, the time to first token (TTFT), i.e., the latency of prefill phase, increases dramatically with the growth on prompt length. In order to concur with such a bottleneck on resources, i.e., long occupation in cloud and limited on-device computing capacity, we propose to separate large language model between cloud and devices. That is, the cloud helps a portion of the content for each device, only in its prefill phase. Specifically, after receiving the first token from the cloud, decoupling with its own prefill, the device responds to the user immediately for a lower TTFT. Then, the following tokens from cloud are presented via a speed controller for smoothed TPOT (the time per output token), until the device catches up with the progress. On-device prefill is then amortized using received tokens while the resource usage in cloud is controlled. Moreover, during cloud prefill, the prompt can be refined, using those intermediate data already generated, to further speed up on-device inference. We implement such a scheme P/D-Device, and confirm its superiority over other alternatives. We further propose an algorithm to decide the best settings. Real-trace experiments show that TTFT decreases at least 60%, maximum TPOT is about tens of milliseconds, and cloud throughput increases by up to 15x.

Aligning Large Language Model Agents with Rational and Moral Preferences: A Supervised Fine-Tuning Approach

Understanding how large language model (LLM) agents behave in strategic interactions is essential as these systems increasingly participate autonomously in economically and morally consequential decisions. We evaluate LLM preferences using canonical economic games, finding substantial deviations from human behavior. Models like GPT-4o show excessive cooperation and limited incentive sensitivity, while reasoning models, such as o3-mini, align more consistently with payoff-maximizing strategies. We propose a supervised fine-tuning pipeline that uses synthetic datasets derived from economic reasoning to align LLM agents with economic preferences, focusing on two stylized preference structures. In the first, utility depends only on individual payoffs (homo economicus), while utility also depends on a notion of Kantian universalizability in the second preference structure (homo moralis). We find that fine-tuning based on small datasets shifts LLM agent behavior toward the corresponding economic agent. We further assess the fine-tuned agents' behavior in two applications: Moral dilemmas involving autonomous vehicles and algorithmic pricing in competitive markets. These examples illustrate how different normative objectives embedded via realizations from structured preference structures can influence market and moral outcomes. This work contributes a replicable, cost-efficient, and economically grounded pipeline to align AI preferences using moral-economic principles.

A Multimodal Deviation Perceiving Framework for Weakly-Supervised Temporal Forgery Localization

Current researches on Deepfake forensics often treat detection as a classification task or temporal forgery localization problem, which are usually restrictive, time-consuming, and challenging to scale for large datasets. To resolve these issues, we present a multimodal deviation perceiving framework for weakly-supervised temporal forgery localization (MDP), which aims to identify temporal partial forged segments using only video-level annotations. The MDP proposes a novel multimodal interaction mechanism (MI) and an extensible deviation perceiving loss to perceive multimodal deviation, which achieves the refined start and end timestamps localization of forged segments. Specifically, MI introduces a temporal property preserving cross-modal attention to measure the relevance between the visual and audio modalities in the probabilistic embedding space. It could identify the inter-modality deviation and construct comprehensive video features for temporal forgery localization. To explore further temporal deviation for weakly-supervised learning, an extensible deviation perceiving loss has been proposed, aiming at enlarging the deviation of adjacent segments of the forged samples and reducing that of genuine samples. Extensive experiments demonstrate the effectiveness of the proposed framework and achieve comparable results to fully-supervised approaches in several evaluation metrics.

Towards Reliable Forgetting: A Survey on Machine Unlearning Verification, Challenges, and Future Directions

With growing demands for privacy protection, security, and legal compliance (e.g., GDPR), machine unlearning has emerged as a critical technique for ensuring the controllability and regulatory alignment of machine learning models. However, a fundamental challenge in this field lies in effectively verifying whether unlearning operations have been successfully and thoroughly executed. Despite a growing body of work on unlearning techniques, verification methodologies remain comparatively underexplored and often fragmented. Existing approaches lack a unified taxonomy and a systematic framework for evaluation. To bridge this gap, this paper presents the first structured survey of machine unlearning verification methods. We propose a taxonomy that organizes current techniques into two principal categories -- behavioral verification and parametric verification -- based on the type of evidence used to assess unlearning fidelity. We examine representative methods within each category, analyze their underlying assumptions, strengths, and limitations, and identify potential vulnerabilities in practical deployment. In closing, we articulate a set of open problems in current verification research, aiming to provide a foundation for developing more robust, efficient, and theoretically grounded unlearning verification mechanisms.

Context-aware TFL: A Universal Context-aware Contrastive Learning Framework for Temporal Forgery Localization

Most research efforts in the multimedia forensics domain have focused on detecting forgery audio-visual content and reached sound achievements. However, these works only consider deepfake detection as a classification task and ignore the case where partial segments of the video are tampered with. Temporal forgery localization (TFL) of small fake audio-visual clips embedded in real videos is still challenging and more in line with realistic application scenarios. To resolve this issue, we propose a universal context-aware contrastive learning framework (UniCaCLF) for TFL. Our approach leverages supervised contrastive learning to discover and identify forged instants by means of anomaly detection, allowing for the precise localization of temporal forged segments. To this end, we propose a novel context-aware perception layer that utilizes a heterogeneous activation operation and an adaptive context updater to construct a context-aware contrastive objective, which enhances the discriminability of forged instant features by contrasting them with genuine instant features in terms of their distances to the global context. An efficient context-aware contrastive coding is introduced to further push the limit of instant feature distinguishability between genuine and forged instants in a supervised sample-by-sample manner, suppressing the cross-sample influence to improve temporal forgery localization performance. Extensive experimental results over five public datasets demonstrate that our proposed UniCaCLF significantly outperforms the state-of-the-art competing algorithms.

Through the Valley: Path to Effective Long CoT Training for Small Language Models

Long chain-of-thought (CoT) supervision has become a common strategy to enhance reasoning in language models. While effective for large models, we identify a phenomenon we call Long CoT Degradation, in which small language models (SLMs; <=3B parameters) trained on limited long CoT data experience significant performance deterioration. Through extensive experiments on the Qwen2.5, LLaMA3 and Gemma3 families, we demonstrate that this degradation is widespread across SLMs. In some settings, models trained on only 8k long CoT examples lose up to 75% of their original performance before fine-tuning. Strikingly, we further observe that for some particularly small models, even training on 220k long CoT examples fails to recover or surpass their original performance prior to fine-tuning. Our analysis attributes this effect to error accumulation: while longer responses increase the capacity for multi-step reasoning, they also amplify the risk of compounding mistakes. Furthermore, we find that Long CoT Degradation may negatively impacts downstream reinforcement learning (RL), although this can be alleviated by sufficiently scaled supervised fine-tuning (SFT). Our findings challenge common assumptions about the benefits of long CoT training for SLMs and offer practical guidance for building more effective small-scale reasoning models.