3447 papers
This paper introduces Directional Decoupling Alignment (D-Align), a novel framework that mitigates Preference Mode Collapse in diffusion reinforcement learning by applying directional corrections to reward signals, thereby preserving generative diversity while achieving superior human preference alignment.
This paper introduces Preguss, a modular framework that combines static analysis with LLM-aided synthesis to automatically generate and refine interprocedural specifications, enabling highly automated verification of large-scale programs (over 1,000 lines of code) while significantly reducing human effort.
To overcome the generalization failure of vanilla CLIP in face forgery detection caused by "low-rank spurious bias," this paper proposes SeLop, a causal representation learning method that identifies and removes spurious correlations via orthogonal low-rank subspace intervention, thereby achieving state-of-the-art performance with high robustness using only 0.39M trainable parameters.
This paper proposes a goal-centric framework for assessing Requirements Engineering methods for Privacy by Design, arguing that practitioners should evaluate these methods based on organizational goals rather than solely on process characteristics to better support their selection and tailoring.
This paper introduces CovertComBench, a specialized benchmark for evaluating Large Language Models in wireless covert communication, revealing that while current models excel at conceptual understanding and code generation, they significantly struggle with the rigorous mathematical derivations required for security-constrained optimization.
This paper proposes a novel AI-generated image detection method that exploits common final architectural components across diverse generators to "contaminate" real images for training, achieving 98.83% average accuracy on unseen generators by leveraging a taxonomy of 21 models and a DINOv3 backbone.
This paper introduces RegionReasoner, a reinforcement learning framework that enforces grounded, multi-round visual reasoning by requiring explicit bounding box citations and global-local semantic consistency, alongside a new benchmark called RegionDial-Bench, to significantly improve spatial grounding and reasoning accuracy in vision-language models.
This paper proves that the conjectured conversion rule, which maps a Rényi Differential Privacy profile to an -Differential Privacy trade-off function via the pointwise maximum of single-order bounds (equivalent to the intersection of RDP privacy regions), is optimal and cannot be uniformly improved upon for any valid RDP profile or Type I error level.
This paper introduces Test-Time Correction (TTC), a training-free method that stabilizes long-sequence video generation in distilled autoregressive models by using the initial frame as a reference anchor to calibrate intermediate states, thereby overcoming error accumulation and extending generation lengths without compromising quality.
This paper introduces \textit{AirPulse}, a 26-gram butterfly-inspired robot that achieves the first autonomous, closed-loop tailless flight at this scale by replicating low-frequency, high-amplitude biomechanical traits through a hierarchical control architecture featuring Stroke Timing Asymmetry Rhythm (STAR).
This paper addresses the issue of modality competition in multimodal learning by theoretically analyzing the relationship between joint and unimodal approaches and proposing TCMax, a hyperparameter-free method that maximizes total correlation between multimodal features and labels to achieve state-of-the-art classification performance.
This survey critically examines the growing body of LGBTQIA+ NLP research within the ACL Anthology, revealing a reactive focus on identifying bias rather than proactive mitigation, and calls for future work to prioritize stakeholder involvement, intersectionality, interdisciplinarity, and non-English languages to build more just and inclusive technologies.
This paper introduces a meta-game framework to evaluate the emergence of algorithmic collusion under test-time constraints by modeling agents with pretrained policies and adaptation rules, revealing how rational choices and co-adaptation influence cooperative or competitive outcomes in repeated pricing games across various algorithmic strategies.
ChimeraLoRA addresses data scarcity in fine-grained tasks by synthesizing diverse and detail-rich images through a hybrid architecture that combines a class-shared LoRA for semantic priors with per-image LoRAs for specific characteristics, guided by semantic boosting and a Dirichlet-based mixture strategy to improve downstream classification accuracy.
DOCFORGE-BENCH introduces the first unified zero-shot benchmark for document forgery detection, revealing that current methods suffer from severe calibration failures due to the extreme rarity of tampered pixels in documents, which renders standard fixed thresholds ineffective and highlights threshold adaptation as the critical missing step for practical deployment.
This paper proposes the Multimodal Adversarial Quality Policy (MAQP), a framework that enhances safe robot grasping in human-robot interaction by introducing a Heterogeneous Dual-Patch Optimization Scheme and a Gradient-Level Modality Balancing Strategy to effectively generate multimodal adversarial patches that address distribution discrepancies and optimization imbalances between RGB and depth modalities.
This paper establishes the first explicit -hardness results for the approximate Covering Radius Problem on lattices in norms for sufficiently large (specifically ), proving that the problem remains hard even for approximation factors approaching $9/8p$ tends to infinity.
This paper presents a production-oriented, unified multilingual semantic retrieval system for Uber Eats that leverages a fine-tuned Qwen2 two-tower model with advanced loss functions and Matryoshka Representation Learning to achieve significant recall improvements across stores, dishes, and grocery items.
This paper introduces the Hybrid Residue Floating Numerical Architecture (HRFNA), a formally verified numerical system combining carry-free residue arithmetic with lightweight exponent scaling that achieves significantly higher throughput, reduced resource usage, and improved energy efficiency on FPGAs compared to IEEE 754 standards while maintaining rigorous, bounded numerical error.
This paper proposes a column-generation-based algorithmic framework to efficiently solve both splittable and unsplittable variants of the capacitated Multi-Commodity Flow problem with convex, potentially non-differentiable, link cost functions, offering a robust optimization approach for managing traffic in telecommunication networks.