All papers

Introduces a framework that dynamically retrieves and recomposes prior latent reasoning steps to guide vision-language-action robot policies.

Transfers editing capabilities from image pairs to video diffusion models using in-context visual demonstrations across decomposed sub-tasks.

Applies zonotope geometry to adaptively scale critic pessimism and disagreement handling in soft actor-critic locomotion policies.

Presents a parameter-free distillation framework transferring knowledge between language models via an iteratively updated natural language guide.

Applies Weisfeiler-Leman graph kernel features to improve automated algorithm selection for constraint optimization problems without training GNNs.

Synthesizes novel views by reusing nearby rendered RGB-D images and applying pose-guided warping with residual refinement.

Evaluates model coherence using 175 specific questions to measure internal contradictions and identity conflation under answer resampling.

Develops an abstraction approach for hierarchical reinforcement learning to balance temporal option boundaries with state-dependent value representation.

Introduces a budget-adaptive tokenization framework that balances token allocation with input length for raw RF signal understanding.

Presents a three-stage hierarchical generation method for large attributed graphs using recursive soft community decomposition.

Proposes a training framework that teaches reasoning models to plan by reverse-engineering solution ideas from hindsight examples.

Unifies robotic policy learning and future state prediction using an autoregressive Transformer trained on spatial sequence tokens.

Deploys a quantized YOLOX-S detector on embedded hardware for onboard maritime satellite imagery selection and data reduction.

Applies confidence sequences to Bayesian neural networks to dynamically allocate Monte Carlo samples per input with statistical guarantees.

Reports an equal-budget study finding randomized-restart list scheduling outperforms Monte Carlo Tree Search on DAG workflow scheduling.

Presents VibeEdit, an image editor conditioned on spatial canvas annotations to perform targeted object additions, modifications, or removals.

Proposes La-Ribo, a geometry-latent flow matching framework designed for simultaneous RNA nucleotide sequence and 3D structure co-design.

Presents AdaCast, a parameter generation method producing input-specific low-rank updates for foundation models in time-series forecasting.

Introduces Cluster-TNN, a mini-batch graph partitioning framework that enables scalable topological deep learning on large graphs.

Benchmarks classical planning against learning-based methods in real-time motion planning under deterministic and stochastic rotating hazard fields.

Audits audio encoders by reconstructing music using a latent diffusion model to quantify signal properties preserved across training objectives.

Proposes Syn-Omni, a framework combining modality-expert LoRA paths and progressive routing to balance shared and modality-specific omnimodal embeddings.

Introduces Dynamic Vector Decoding to efficiently map 2D and 3D perception coordinates into compact discrete tokens for multimodal LLMs.

Surveys recent developments in physics-integrated machine learning applications across high-energy nuclear physics experiments and simulations.

Introduces grasp, a framework combining functional gradient boosting with LLM proposal generation to learn probabilistic logic programs from relational data.

Analyzes attention-head activation graphs to show that small-world connectivity correlates with reasoning performance in large language models.

Proposes a formal structural framework defining context formation, abstraction, and invariance operations for cognitive representation and problem solving.

Evaluates whether LLM compliance systems adjust verdicts when governing rules are perturbed, finding many verdicts are insensitive to rule changes.

Develops a framework fusing small randomized controlled trials with large observational data to improve treatment effect estimation precision without bias.

Presents RiCo, a neural physics simulation method that models rigid-body interactions via sparse local contact surface reasoning.

Proposes ContiLNN, which uses closed-form continuous-time modules to improve cross-slice continuity and restoration in 2D medical imaging.

Analyzes long-tail concept learning during supervised fine-tuning using a theoretical prior barrier metric to guide data needs.

Distills geometric privilege into vision-language models during training to improve spatial reasoning without inference overhead.

Audits legal chain-of-thought faithfulness by substituting cited authorities and observing whether model verdicts causally change.

Introduces an annotated dataset of handwritten student answer sheets with complex formulas, text, tables, and noise artifacts.

Presents a method to improve zero-shot robot foundation model performance using causal reasoning traces generated from demonstrations.

Proposes an evolutionary framework that optimizes geometric diagram formalizations to improve multimodal geometry problem solving.

Proposes a higher-order tensor modeling framework to preserve synergistic cross-modal interactions in self-supervised multimodal representations.

Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives.

Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning.

Applies psychometric auditing to HELM Safety benchmarks to test whether harmful refusal functions as a single measurable attribute.

Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values.

Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis.

Develops multi-layer linear probes on model activations to detect agent deception and sabotage across diverse scenarios.

Proposes a generative flow matching method with repulsion-guided sampling to model one-to-many bifurcation trajectories in physical systems.

Recomputes METR task time horizons using splines and item-response theory to examine the validity of AI difficulty curves.

Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations.

Investigates self-preservation in language models by using task banks to steer synthetic replay during continual instruction tuning.

Proposes a mesh-native tokenization approach to adapt foundation model architectures to unstructured scientific simulation data.

Benchmarks machine learning and pretrained foundation models for day-ahead electricity forecasting in a university building.

Demonstrates a backdoor attack where malicious reasoning degradation is triggered incrementally using invisible prompt tokens and stage tracking.

Introduces a key-value cache compression method representing attention activations as sums of compact Kronecker products.

Analyzes zero-advantage training plateaus in reinforcement learning with verifiable rewards and proposes recursive data synthesis to address them.

Evaluates an incremental retrieval-augmented generation framework that progressively populates its knowledge base from human decision precedents.

Introduces a benchmark evaluating language models on eliciting geospatial data requirements and constraints from natural language requests.

Trains an unsupervised vision-language reward model using intermediate layer representations and caption perturbations without human preference labels.

Develops a temporal encoder injecting engineering domain priors to forecast rare, high-cost events in multivariate time series.

Constructs a multimodal simulator dataset and trajectory prediction model for autonomous driving in urban roadway work zones.

Introduces an automated pipeline using language models to construct hierarchical knowledge graphs from complex hardware datasheets.

Presents a certified semantic filtering technique that reuses prior model labels to reduce language model calls in database queries.