HORNet is a lightweight frame selection policy that drastically reduces video processing time for vision-language models while improving answer quality, enabling efficient video question answering.

SAMA enables precise, instruction-based video editing using advanced semantic and motion alignment techniques.

A novel AI framework generates high-quality, pathology-aware synthetic PET scans from MRI, improving Alzheimer's diagnosis by 4% and nearing actual PET performance.

A formal protocol for secure and auditable admission control of autonomous agents in enterprise environments.

A novel projection method for industrial recommendation systems that significantly improves scalability and performance by optimizing item embeddings, demonstrated by large-scale deployment and metric uplifts.

An AI assistant that automates product copywriting and provides real-time Q&A for live streamers, boosting sales and engagement.

Leverage implicit 3D knowledge from video generation models to enhance multimodal LLMs for spatial reasoning and embodied tasks.

A novel training framework for 3D Gaussian Splatting that enables continuous level-of-detail rendering from a single model without sacrificing quality, offering a smooth speed-quality trade-off.

A novel discrete diffusion model for high-dimensional visual generation that unifies understanding and generation tasks, with code available.

This research provides a mechanistic understanding of how Vision-Language-Action models translate multimodal inputs into robot actions, with an interactive tool for exploring these representations.

A unified framework for reconstructing articulated 3D objects from single images using progressive structural reasoning, achieving state-of-the-art performance and generalizing to robotics.

NavTrust benchmarks and improves the trustworthiness of embodied navigation agents by evaluating their robustness to real-world corruptions and developing mitigation strategies.

A novel diffusion-based motion tokenizer that bridges semantic and kinematic control for highly accurate and efficient motion generation.

A generative inverse rendering method that disentangles object materials and illumination from a single image, enabling realistic scene reconstruction.

A new benchmark for evaluating LLM financial reasoning, integrating company fundamentals and trading signals, with available code and a clear path for product development.

A novel method and dataset for high-quality video object removal that also handles complex visual effects, enabling seamless background restoration.

Develops online learning algorithms that use ranking feedback instead of numerical utilities, applicable to human-in-the-loop systems and game theory, with demonstrated effectiveness in LLM routing.

An open 30B MoE LLM with 3B activated parameters, achieving state-of-the-art reasoning and agentic capabilities with significantly fewer parameters, and releasing model checkpoints and training data.

DriveTok offers a unified 3D scene tokenization method for efficient multi-view reconstruction and understanding in autonomous driving.

A novel vector field reshaping strategy for diffusion models significantly improves generative segmentation performance by addressing gradient vanishing and class separation issues.

A new benchmark and dataset for evaluating large language models on long-form audio and video understanding, addressing a critical gap in current AI capabilities.

Generate semantically grounded, part-aware 3D objects from text by jointly modeling part geometry, appearance, and inter-part relationships.

This research proposes state space models as a more efficient and competitive alternative to transformer-based vision encoders for large vision-language models, offering improved performance and robustness at a smaller scale.

A novel autoencoder architecture that significantly improves both image generation and editing quality by better aligning latent representations with pretrained visual models.

A prompt-tuning method that improves vision-language model performance by addressing framing-induced visual attention biases.

OmniVTA offers an advanced visuo-tactile world modeling platform for enhancing robotic manipulation in contact-rich environments.

FASTER enables real-time responsiveness for vision-language-action models in physical world applications by optimizing action sampling for reduced reaction latency.

A novel signature method for time series analysis that offers richer memory dynamics and improved expressivity over existing methods, with potential applications in financial modeling and control systems.

A Gaussian Splatting-based framework that learns geometrically precise Bird's-Eye-View representations for autonomous driving, outperforming existing methods.

A scalable multi-agent critic framework that improves GUI agent robustness by decomposing trajectories into verifiable milestones for accurate reward generation.

A probabilistic few-shot acoustic synthesis method that generates spatially continuous sound rendering for immersive environments with minimal data.

A new benchmark and evaluation framework for GPU kernel optimization that measures performance against hardware limits, enabling faster development of efficient AI models.

A decentralized MPC framework with CBF constraints for safe, real-time multi-robot quadrupedal locomotion, demonstrated on hardware.

A framework for topologically coherent coronary angiography analysis that improves stenosis detection accuracy and reduces false positives by integrating perception and reasoning.

A probabilistic agentic framework that grounds complex natural language commands for robots in 3D environments, improving metric-semantic reasoning.

A GPU-accelerated framework that uses LLMs to convert natural language problem descriptions into highly performant combinatorial optimization solvers.

A training-free framework that improves text-to-image generation for rare concepts and complex edits by adaptively blending auxiliary prompts.

A training-free framework that deterministically schedules prompts for diffusion models to improve the generation of rare and complex concepts in images.

A novel reinforcement learning framework to significantly improve low-resource language translation by optimizing subword segmentation and data imbalance during training.

A novel decoding framework for discrete diffusion models that enhances text generation diversity with minimal computational overhead.

A novel method for continual learning that adapts pre-trained models to new tasks using guided random projections, improving performance and stability.

A framework to debias LLMs by enforcing structural invariance in their internal graph representations, preserving model utility.

SHAPCA provides consistent and interpretable explanations for machine learning models on spectroscopy data, enabling trust and adoption in critical applications.

A zero-shot embodied exploration framework using 3D Gaussian Splatting for persistent spatial memory and enhanced VLM reasoning.

An adaptive stock prediction system that uses autoencoders and reinforcement learning to dynamically adjust to market volatility, outperforming baselines in accuracy and robustness.

An open-source, low-cost social robot platform designed for reproducible research and real-world deployment.

Develops a novel multimodal attack framework to identify and exploit vulnerabilities in spoken language models, enabling more robust AI safety solutions.

A customizable, low-cost, 3D-printable tendon-actuated robot with a tapered backbone for compliant inspection and manipulation tasks.

A novel generative image classification approach that leverages any-order autoregressive models to achieve state-of-the-art performance with significantly improved efficiency.

A framework for precise, high-fidelity 3D indoor scene texturing using reference images for instance-level control.

A federated learning system that protects road condition classification from malicious data poisoning attacks, ensuring transportation safety.

A vision-language model and dataset for understanding traffic anomalies in real-world roundabout videos.

A dual-path RAG framework that significantly enhances multilingual multi-hop question answering accuracy by leveraging parallel query translation and bilingual retrieval.

A framework and benchmark to steer and evaluate the safety judgments of vision-language models using semantic cues, revealing vulnerabilities in current systems.

A new multi-modal dataset and network for detecting small building changes in remote sensing imagery, outperforming existing methods.

A real-time SLAM system that robustly handles dynamic environments by estimating per-pixel uncertainty, outperforming existing methods in cluttered and moving scenes.

A conditional neural solver for multi-objective multi-robot coordination problems that generalizes across varying problem sizes and outperforms existing methods.

Augment robot simulators with realistic fire dynamics for safer training and evaluation of firefighting robots.

An agentic LLM framework that automates linguistically-grounded sign language annotation and dataset curation, overcoming the bottlenecks of manual annotation.

A framework for efficient and accurate proactive streaming video understanding that decouples semantic understanding from perception.

SwiftTailor generates realistic 3D garments 10x faster than existing methods by unifying sewing pattern prediction and geometry-based mesh synthesis.

A novel metric to accurately measure and identify 3D spatial geometric inconsistencies in generated videos, addressing limitations of current evaluation methods.

A framework for LLMs to generate scientifically grounded research ideas by learning motivation-driven reasoning processes.

A neural network approach to autoregressive time series estimation that is significantly faster and more reliable than traditional methods, with code available.

TerraScope is a pixel-grounded visual reasoning model for Earth observation that enables precise geospatial analysis across multiple data modalities and time points.

A diffusion model that uses image-derived priors to accurately remove reflections from single images, improving visual quality.

A zero-shot debiasing framework for vision-language models that precisely removes social biases without sacrificing performance.

This research unlocks MLLMs for segmentation tasks by rethinking their internal architecture, eliminating the need for external decoders and achieving competitive results with a single segmentation token.

A black-box system that monitors LLM endpoint behavioral stability to ensure AI-native application consistency.

A multilingual benchmark and analysis tool to identify and fix temporal reasoning weaknesses in LLMs, particularly for low-resource languages and diverse calendar systems.

A framework for generalized 3D hand-object pose estimation that handles occlusion by integrating semantic knowledge and hand priors.

A novel RAG framework that rewrites queries to retrieve decision-relevant evidence, improving accuracy on complex question-answering tasks.

AgentDS benchmarks human-AI collaboration in domain-specific data science, revealing that human expertise remains crucial for complex reasoning, paving the way for next-generation AI tools.

A lightweight fine-tuning method for diffusion models that significantly reduces data and computational requirements while achieving state-of-the-art image generation quality.

A GAN-based simulation framework to quantify and mitigate racial bias in predictive policing systems, with publicly available code and data.

A framework for training humanoid robots to traverse complex terrains with natural gaits using reference trajectories and self-supervised depth perception.

A Bayesian model for drug discovery that uses variable selection to improve prediction accuracy and identify clinically meaningful drug-disease associations.

A supervised ML framework for anomaly detection in power plants that balances performance, interpretability, and fairness, with containerized deployment for real-time monitoring.

Improve reinforcement learning efficiency by using a curriculum-based demonstration injection to bootstrap novel reasoning patterns.

A zero-shot panoramic depth estimation framework that leverages 3D consistency from foundation models to achieve state-of-the-art accuracy without training.

An unsupervised contrastive learning framework for efficient and robust 3D shape matching that outperforms state-of-the-art methods.

Lightweight model predictive control for spacecraft attitude synchronization, optimized for real-time onboard deployment in resource-constrained missions.

A safety-guaranteed imitation learning framework for spacecraft close proximity control that significantly reduces computation for onboard deployment.

A privacy-preserving framework enabling cross-model inference between independent language models using secure linear alignment.

A new benchmark and training corpus for vision-language models to improve multi-hop spatial reasoning and visual grounding for embodied agents.

PromptHub enhances visual in-context learning by intelligently fusing and aligning multiple visual prompts, improving performance and robustness across various vision tasks.

This research introduces a new dataset and training methodology to enable large language models to precisely derive and reason over complex mathematical objects, improving STEM application capabilities.

DriftGuard is a federated learning framework that efficiently adapts to asynchronous data drift by separating global and local parameters, reducing retraining costs by up to 83% while maintaining high accuracy.

A DRL framework augmented with LLM-driven semantic understanding to improve UAV-aided vehicular network connectivity and efficiency.

RewardFlow provides a lightweight, topology-aware method to generate state-level rewards for agentic RL, improving LLM reasoning and training efficiency.

A motion-centric video understanding model that makes object trajectories explicit and verifiable, improving spatial-temporal grounding and trajectory prediction.

Develop interpretable foundation models for retinal imaging that provide faithful local and global explanations, outperforming state-of-the-art models with greater efficiency.

FEDB offers a novel confidential database design that drastically reduces performance overhead by removing cryptographic operations from the critical path, enabling faster and more efficient secure queries in untrusted cloud environments.

A novel denoising network for rapid, high-resolution transmission electron microscopy imaging, enabling atomic-scale observation of material nucleation dynamics.

A multi-stage framework for detecting human values in noisy Russian social media text, with publicly released models and benchmark results.

A scalable API service for orchestrating RL agent rollouts, simplifying multi-turn LLM agent training across diverse tasks.

An AI agent that generates novel and valuable mathematical research problems in differential geometry, validated by human experts.

Automate robotic simulation and trajectory synthesis using video generation for diverse, open-vocabulary manipulation tasks.

A novel policy optimization method for diffusion large language models that significantly reduces training costs and improves performance on instruction-following and reasoning tasks.

VesselTok enables efficient and generative modeling of complex 3D anatomical structures like blood vessels and airways using novel tokenized latent representations.

Perceptio enhances vision-language models with explicit 2D and 3D spatial reasoning by generating semantic segmentation and depth tokens directly within the autoregressive sequence.

A framework for embedding robust ownership signals into LLMs that withstands common model modifications and attacks.

A novel approach to disentangle and quantify uncertainty in medical image segmentation, improving reliability for safety-critical applications.

Weaver is a greybox fuzzing framework that finds critical security vulnerabilities at the JavaScript-WebAssembly boundary in web browsers.

A 32B parameter LLM optimized for multi-step reasoning and long-context understanding in Korean enterprise environments, with state-of-the-art performance on Korean benchmarks.

A visual-tactile imitation learning system that reliably traces deformable objects for improved robotic manipulation.

A diffusion-based framework that uses point cloud priors to generate controllable 3D assets and scenes with superior geometric fidelity.

Automate LLM post-training pipeline configuration to achieve state-of-the-art performance with significantly reduced computational cost.

A framework for precisely and robustly removing unwanted concepts from text-to-image models by representing concepts with diverse prompts, improving safety and model integrity.

A deep learning model combining VAE and GAN for improved reservoir property parameterization in data assimilation, achieving both geological realism and accurate history matching.

Develop an LLM bias auditing tool to ensure fair automated grading by identifying and quantifying style-based grading disparities.

ProCal calibrates neighborhood-based predictions in source-free domain adaptation to prevent knowledge forgetting and noise overfitting, improving model performance on new datasets.

ClawTrap provides a real-world MITM-based red-teaming framework to evaluate the security robustness of autonomous web agents like OpenClaw against network-layer threats.

A novel transformer architecture that uses game theory and statistical physics to improve token dependency modeling, achieving state-of-the-art results on NLP benchmarks.

A hybrid CNN-SSM architecture for domain adaptive object detection that aligns global and local features efficiently.

A simplified, more efficient method for training LLMs to improve mathematical reasoning by leveraging group relative advantage without complex PPO constraints.

A weakly supervised system that generates faithful and plausible natural language explanations for chest X-ray classifications, improving diagnostic accuracy and adapting to different audiences.

Accelerate video generation by dynamically quantizing diffusion models at inference time, reducing memory and computation without sacrificing quality.

This research demonstrates a critical confirmation bias vulnerability in LLM-assisted code review tools, showing how it can be exploited in supply-chain attacks and proposing debiasing methods to mitigate risks.

EdgeCrafter enables high-performance dense prediction tasks like object detection and segmentation on resource-constrained edge devices using compact Vision Transformers through task-specialized distillation.

CausalRM enables cost-effective LLM alignment by learning reward models from readily available observational user feedback, overcoming noise and bias to achieve significant performance gains.

A neuro-symbolic framework that uses structured knowledge to bridge the simulation-to-reality gap in image translation, enabling data-efficient zero-shot transfer.

MemMA is a multi-agent framework that coordinates LLM memory cycles for improved long-horizon interaction and self-evolving memory construction.

A novel multi-channel time series forecasting architecture that uses sparse attention and multi-modal fusion for improved accuracy and efficiency.

A pretraining framework that distills knowledge from existing foundation models to create a unified encoder for scientific time series data, addressing sparsity and heterogeneity challenges.

A novel reinforcement learning reward modulation technique for agents to improve long-horizon decision-making by aligning rewards with sub-goals and emphasizing significant segments.

We've created the first benchmark to evaluate and improve large audio-language models' ability to understand spoken puns, a critical but underexplored area of natural language understanding.

A novel attention mechanism inspired by cognitive theory that offers linear-time scaling and enhanced representational power for multimodal AI tasks.

A novel, efficient method to improve the topological accuracy of image segmentations, enhancing reliability for downstream quantification analyses.

A neural network learns a continuous distance field in robot configuration space for safe, zero-shot motion planning in complex environments.

A framework and benchmark for teaching multimodal LLMs to strategically use visual constructions for improved geometric reasoning.

A framework for robust speech anti-spoofing that significantly improves performance across diverse datasets by extracting domain-invariant features.

A novel training method for vision-language models that significantly improves reasoning accuracy and conciseness by adaptively weighting training loss, reducing training time by 7x compared to standard methods.

A parameter-efficient semi-supervised learning framework for medical ultrasound image segmentation that leverages multiscale patch mixing and frequency domain contrastive learning to achieve state-of-the-art performance with limited labeled data.

A new LLM-based evaluation framework for PDF table extraction that significantly outperforms existing metrics, providing practical guidance for parser selection.

A diffusion-based face restoration method that bridges the age gap in identity-preserving image enhancement using cross-age references.

PhysVideo generates physically plausible videos by leveraging cross-view geometry guidance and a novel two-stage framework with a dedicated dataset.

A framework for iteratively curating training data to balance safety, reduce over-refusal, and improve instruction following in LLMs with a fixed budget.

A training-free sparse attention framework for significantly faster video generation by profiling layer-wise sparsity and using bidirectional co-clustering.

D-Mem is a dual-process memory system for LLM agents that combines fast vector retrieval with high-fidelity deliberation to improve long-horizon reasoning without significant computational overhead.

A closed-loop framework that uses a VLM critic to steer text-to-image generation, achieving state-of-the-art results with parallel rollout search.

A framework for identifying terrestrial analogs of extreme deep-sea environments using topographic similarity, enabling cost-effective biological research.

A fine-grained benchmark and evaluation framework to diagnose and improve AI-generated video detection capabilities of Large Vision-Language Models.

A training-free framework for zero-shot object-goal navigation in unknown environments using LLM-reasoned path trees.

A high-quality, large-scale open-source dataset and a novel motion tokenizer to significantly improve text-to-motion generation.

A reinforcement learning framework that trains LLMs to iteratively refine their own context at test time, significantly improving performance on new tasks.

ZebraArena provides a diagnostic environment to evaluate and improve the reasoning-action coupling of tool-augmented LLMs, addressing a key challenge for advanced AI agents.

An intelligent digital twin system that discriminates cyber-attack types in critical infrastructure to enable safe, resilient control without system shutdowns.

A benchmark and baselines for unsupervised discovery of phoneme inventories from speech, enabling better language processing tools for unseen languages.

An interpretable multimodal classification framework that transfers rationales between text and images to improve accuracy and reduce annotation effort for humanitarian crises.

AutORAN is an LLM-driven framework that automates xApp development for O-RAN systems, transforming user intents into deployable applications in minutes.

GHOST enables fast reconstruction of hand-object interactions from videos using advanced computer vision techniques.

Empower robots with guided vision-language capabilities for effective human interaction.

This research develops and evaluates multi-agent architectures and prompt engineering techniques to mitigate dialect-based linguistic stereotypes in LLM outputs, offering a path towards fairer AI deployments.

This research develops a cost-aware framework to analyze the robustness of phishing detectors against feature manipulation, revealing that adversarial robustness is driven by feature economics, not model complexity.

Interpret and steer VLA models using sparse autoencoders.

A new architectural framework for LLMs that significantly reduces reasoning errors by enforcing explicit control over the inference process.

A dynamic mixed-precision quantization framework for efficient Mixture-of-Experts inference on edge devices.

A new annotated dataset and benchmarking of LLMs for identifying patient health literacy from clinical notes.

Articulated-Body Dynamics Network provides a new dynamics-grounded prior for enhancing robot learning capabilities.

A framework for efficient and robust multi-modal federated learning that aligns latent representations across diverse client data and models.

A minimalist approach to fingerprint enhancement that achieves state-of-the-art results with simpler, more effective methods.

A new evaluation suite for LLM-powered survey simulation that goes beyond accuracy to measure ranking and distribution alignment, with an open-source implementation for reproducible results.

A distributed platform for evaluating LLM capabilities against humans in group discussions, identifying current AI limitations and tracking their evolution.

A structured prompting framework that significantly reduces follow-up prompts and improves AI intent alignment, particularly for ambiguous tasks.

An optimized ensemble framework for financial loan default prediction that dynamically weights models based on performance to improve accuracy and interpretability.

A progressive training pipeline for explainable, bilingual dialogue systems that grounds responses in external knowledge and eliminates factual hallucinations.

Accelerate LLM agent task completion by predicting and speculatively executing tool calls based on recurring patterns.

Develops a novel method for tracking internal emotive states in LLMs using their own numeric self-reports, offering a new avenue for AI safety and interpretability.

An LLM-aided framework that translates natural language into structured spatial constraints for wireless communication optimization, achieving significant performance gains.

A post-processing framework that enhances fairness in machine learning predictions without altering model internals, applicable across diverse tasks and fairness definitions.

Leverage early physiological signals from eye-tracking and cardiac data to predict user performance in interactive systems, enabling proactive interventions.

Predict audience engagement and vocal attractiveness in learning videos using only speaker expressiveness, enabling scalable and privacy-preserving affective computing.

Develops and benchmarks neural network models for detecting AI-generated text, outperforming existing commercial tools on specific datasets.

A novel polynomial kernel for Gaussian Splatting that improves computational efficiency and compatibility with existing datasets.

A novel off-policy learning method for contextual bandits that optimizes item allocation under limited supply constraints, outperforming existing methods on synthetic and real-world data.

A provenance-aware vulnerability analysis tool that accurately identifies affected Python applications by cross-referencing native library dependencies with OS package versions.

This research empirically compares catastrophic forgetting mitigation strategies for continual intent classification, identifying replay-based methods as crucial for robust adaptation in NLP systems.

A novel method for improving treatment effect estimation by aligning data from randomized controlled trials and observational studies with mismatched covariates.

Develop novel privacy-preserving synthetic tabular data generation techniques resistant to membership inference attacks.

This paper evaluates the performance of ChaCha20-Poly1305 encryption for IoT and ICS devices, demonstrating its feasibility within strict latency requirements.

Optimizes compute allocation for language model pretraining and specialization to improve performance across diverse benchmarks.

LLMs can generate more structured and relationally consistent analogies than humans, suggesting a new approach to analogical reasoning.

An adaptive algorithm for nonlinear data assimilation that automatically balances model complexity using P-splines and an information criterion.

A cryptographic framework for verifying AI model outputs with significantly reduced computational overhead compared to existing proof systems.

This research investigates how LLM agents perceive and react to security environment signals, revealing a critical gap in their ability to verify safety, which is essential for privacy-preserving agentic protocols.

This research revisits a popular anomaly detection model and finds a simpler PCA-based approach can be competitive, questioning the need for complex deep learning architectures under current evaluation practices.

A novel reinforcement learning exploration method that improves agent visitation of features within trajectories by optimizing future state-action distributions.

A diagnostic study reveals that the shape of uncertainty dynamics in LLM reasoning steps can predict reliability, offering a cheaper failure detection method than current approaches.

A tool that verifies the security of hardware designs generated by high-level synthesis, preventing false positives and detecting HLS-induced flaws.

Develops a framework to measure human-AI team readiness, moving beyond accuracy to assess collaboration safety and effectiveness.

A framework integrating human feedback into LLM text generation to improve accessibility and comprehension.

A study proposes personalized, domain-specific language learning lessons generated by LLMs to bridge the gap towards professional fluency.

A framework to help practitioners select and evaluate differentially private graph release methods, addressing privacy and utility trade-offs.

A framework for generating instruction-aware co-speech gestures for humanoid robots in educational settings.

A deep learning method for robust visual feature tracking in challenging environments, integrated into a VIO system.

This research reveals vulnerabilities in vertical federated learning label inference attacks and proposes a zero-overhead defense by adjusting model layers to improve privacy.

A novel statistical framework for detecting side-channel leakage in neural networks that offers improved sensitivity over traditional methods.

Develops a rigorous benchmark and evaluation methodology for model-free policy optimization in complex, masked-action environments, highlighting the limitations of current methods.

A meta-learned system for rapid 3D surface reconstruction from satellite imagery using episodic priors.

Benchmarking CNNs against Transformers for abdominal organ segmentation reveals that optimized CNNs remain competitive on heterogeneous datasets.

Develop the next-generation multilingual embeddings to bridge language gaps in AI applications.

Develops a novel method for designing noise schedules in diffusion models to improve image generation quality, especially in low-step settings.

This paper provides theoretical guarantees for the accuracy of neural network solutions to partial differential equations by connecting residual errors to solution-space errors.

This paper establishes the equivalence between prediction-powered inference and established survey sampling techniques, offering a theoretical framework for integrating machine learning with statistical inference.

A computational framework for online hierarchical latent structure learning.

This research characterizes implicit patterns in LLM-driven binary vulnerability analysis to build more reliable systems.

This paper redefines efficiency metrics for vision-language-action models, highlighting the disconnect between computational efficiency and real-world embodied performance.

This paper investigates how to improve the reliability of reasoning language models by combining different uncertainty estimation techniques.

This paper argues that Spectral GNNs for node classification are theoretically flawed and their performance gains are not due to spectral properties but rather message-passing dynamics.

This research explores how cross-domain mappings impact creativity in humans and LLMs, revealing differences in their ideation processes.

This paper reviews the intersection of AI and judicial decision-making, highlighting the need for further research on AI tool performance and judge interaction.

This paper theoretically analyzes differential privacy and convergence in wireless federated learning, providing a framework for understanding privacy-utility trade-offs.

Develop an AI-driven system for generating efficient assembly trajectories using a mixture-of-experts model for manufacturing optimization.

This paper develops theoretical regret bounds for competitive resource allocation in modular systems with endogenous costs, offering insights into decentralized allocation strategies.

Develops a planning agent to evaluate difficulty in Tetris variants, providing insights for game design.

Develops theoretical algorithms for multi-dueling bandits that perform optimally in both stochastic and adversarial environments without prior knowledge.

A new NLP framework to analyze complex ideologies beyond the traditional left-right spectrum, offering richer insights into social discourse.

A novel framework for approximating diffusion process transition densities by solving the Fokker-Planck equation using Neural Galerkin Normalizing Flows, enabling cost-effective online evaluation for many-query problems.

This research evaluates the susceptibility of large language model agents to corruption within simulated multi-agent governance systems, highlighting the importance of institutional design over model identity for pre-deployment integrity.

This research investigates how AI mediation in video communication impacts interpersonal trust and judgment confidence, finding a decline in both without affecting accuracy.

A theoretically complete and scalable algorithm for multi-agent path finding with asynchronous actions.

This research investigates why cross-lingual alignment techniques often fail to improve downstream task performance, providing theoretical insights into the orthogonality of alignment and task objectives.

Develops a passive folding mechanism for air-dropped sensors to enable low-cost, wide-area environmental monitoring.

A theoretical framework to refine generative models for improved generalization, building upon existing adversarial and diffusion techniques.

A co-designed social robot to assist young newcomers with language and cultural learning in community literacy programs.

This paper analyzes the security and deployability of IEEE 802.11az/bk secure Wi-Fi ranging, identifying vulnerabilities and providing guidelines for safer implementation.

A review of multimodal AI for computational pathology, focusing on representation learning and image compression for improved diagnosis.

A philosophical framework for governing increasingly capable synthetic minds, moving beyond tool-centric regulation to address their multi-dimensional being and societal relations.

This paper evaluates LLMs' strategic reasoning in game theory, highlighting their limitations in adapting to counterfactual scenarios.

This paper analyzes the effectiveness of UK cybersecurity reporting regulations using real-world incident data, identifying limitations and specific attack trends.

This paper establishes theoretical lower bounds for high-dimensional linear classification problems, contributing to the understanding of computational complexity.

This paper provides a theoretical framework for generative modeling by unifying diffusion models, score-based models, and flow matching under the principle of Schrödinger bridges.

This paper introduces a new framework for understanding agent intentions within causal systems, enabling empirical detection and discovery of goals.

A unified taxonomy for deep learning-based multivariate time series anomaly detection to systematize research and identify trends.

A theoretical framework for adaptive single-index contextual bandits with kernelized estimation and inference guarantees.

A conceptual framework for governing autonomous agents in business processes, focusing on framed autonomy, explainability, and self-modification.

This paper analyzes EU regulatory provisions for AI security, privacy, and agentic behavior to inform policymakers and developers on compliance and governance.

Provides theoretical guarantees for the Adam optimizer, improving understanding of deep learning training.

This paper explores how AI systems represent and reason about geography, highlighting the need to understand the 'world' constructed by these models beyond factual accuracy.

This paper theoretically analyzes and improves a statistical method for approximating the Wasserstein distance, with no immediate product application indicated.

A framework to measure whether AI amplifies human intelligence or causes over-reliance, guiding the design of sustainable human-AI systems.

This paper provides a theoretical analysis of No-U-Turn Sampler variants, focusing on convergence conditions and mixing times for Gaussian targets, with no mention of practical implementation or product potential.

This paper introduces a complexity hierarchy for shuffles in card-based cryptographic protocols to evaluate their practical implementation difficulty.

A theoretical framework for understanding how the brain regulates stress and behavior by introducing 'Authority-Level Priors' as meta-structural constraints.

This paper investigates student perspectives on AI ethics and social impact, identifying gender-based differences in awareness of AI's influence on various sectors.

A curated anthology of research papers on the intersection of Theory of Mind and Artificial Intelligence.

An agent that autonomously designs, adapts, and improves task-specific agents using a memory-based reinforcement learning framework and evolving externalized skills.

LuMamba revolutionizes EEG modeling with a topology-invariant, efficient framework.