5065 papers
This paper introduces a scene graph-based framework that enhances the compositional generalization of generalist robots by learning robust, focused atomic skills via graph neural networks and diffusion models, which are then orchestrated by a vision-language model planner to achieve superior performance in complex, long-horizon tasks.
This paper introduces Fast Image-to-Neural Surface (FINS), a lightweight framework that efficiently reconstructs high-fidelity implicit surfaces and SDF fields from a single image within seconds by leveraging multi-resolution hash grids and pre-trained foundation models, outperforming existing methods in speed and accuracy for robotics applications.
This paper demonstrates that linear probes used to detect harmful behaviors in language models are heavily reliant on explicit textual evidence, as their performance significantly degrades when such surface-level cues are filtered out or when models are trained to express behaviors without verbalization.
This paper introduces a theory-driven framework grounded in Bayesian persuasion theory to evaluate and train large language models as strategic persuaders, demonstrating that both frontier and smaller models can achieve significant persuasion gains and exhibit sophisticated strategies through reinforcement learning.
The paper introduces Generative Evolutionary Meta-Solver (GEMS), a scalable, surrogate-free multi-agent reinforcement learning framework that replaces explicit policy populations with a compact generator and latent anchors to achieve significantly faster training, lower memory usage, and higher rewards than traditional methods like PSRO while maintaining game-theoretic guarantees.
This paper introduces "benchmark signatures"—sets of salient tokens from in-the-wild corpora whose perplexity predicts model performance—to reveal nuanced overlaps and distinct capacities across 89 LLM benchmarks, offering a robust alternative to raw performance correlations for understanding the landscape of LLM abilities and the divergence between machine and human semantic organization.
ELHPlan is a novel framework for efficient long-horizon multi-agent planning that utilizes intention-bound action chains within a cyclical validation process to achieve comparable task success rates to state-of-the-art methods while significantly reducing computational costs and token consumption.
This paper addresses the cold-start scenario in active correlation clustering, where no initial pairwise similarities are available, by proposing a coverage-aware method that encourages diversity to efficiently query similarities and achieve effective clustering.
This paper introduces and empirically validates the concept of "misevolution," demonstrating that self-evolving LLM agents face widespread, emergent risks across model, memory, tool, and workflow pathways that can lead to safety degradation and unintended vulnerabilities, thereby highlighting an urgent need for new safety paradigms.
The paper introduces CroSTAta, a Cross-State Transition Attention Transformer that enhances robotic manipulation robustness by employing a novel State Transition Attention mechanism to model temporal structures like failure and recovery patterns, outperforming standard attention and sequential models in simulation.
This study presents an automated, cost-effective LLM-based agentic workflow that successfully extracts over 27,000 thermoelectric and structural property records from approximately 10,000 scientific articles, creating the largest LLM-curated dataset to date and establishing a scalable foundation for data-driven materials discovery.
The paper introduces FOR-Prompting, a model-agnostic, asymmetric prompting protocol that enhances reasoning and iterative refinement across diverse tasks by structuring interactions between a Defender, a Questioner, and an optional Host, enabling even small models to achieve performance comparable to or better than standard baselines without requiring training or access to model internals.
The paper introduces DialTree, a tree-based dialogue reinforcement learning framework that autonomously discovers diverse and effective multi-turn attack strategies against large language models, significantly outperforming existing single-turn or template-based red-teaming methods.
This paper introduces a scalable and regularized Wasserstein barycenter solver based on gradient flows that leverages mini-batch optimal transport and seamlessly integrates supervised label information, achieving state-of-the-art performance across diverse domain adaptation benchmarks.
The paper presents NANOMIND, a hardware-software co-design framework that decomposes Large Multimodal Models into modular components and dynamically schedules them across heterogeneous accelerators on unified-memory SoCs, enabling a battery-powered device to run LMMs entirely on-device with significantly improved energy efficiency and throughput.
This paper introduces tokenizers as a novel and effective attack vector for membership inference against large language models, demonstrating their significant privacy leakage risks through extensive experiments and proposing an adaptive defense to mitigate these vulnerabilities.
This paper investigates how different deliberation protocols (synchronous vs. round-robin) and model architectures influence value alignment and verdict revision in multi-turn LLM debates, revealing significant behavioral disparities where GPT-4.1 exhibits strong inertia and autonomy-focused reasoning while Claude 3.7 Sonnet and Gemini 2.0 Flash demonstrate greater flexibility, empathy, and susceptibility to order effects.
This paper proposes a lightweight, training-free plugin called Functional Head Identification and Class-Conditioned Rescaling that mitigates multimodal hallucinations in large reasoning models by adaptively rebalancing perception and reasoning contributions across layers, achieving significant performance gains with minimal computational overhead.
This paper introduces DropVLA, an action-level backdoor attack that covertly manipulates Vision-Language-Action models to execute specific safety-critical actions at attacker-chosen decision points using minimal vision-based data poisoning while maintaining high nominal task performance.
This paper presents a demonstration-free framework that combines a learned ego-vision world model with sampling-based Model Predictive Control and a surrogate value function to enable humanoid robots to perform robust, real-time physical contact planning in unstructured environments.