Test-time adaptation approach for improving deep learning model generalization across geographic regions in remote sensing land surface temperature prediction.
Framework using non-equilibrium stochastic dynamics to address stability-plasticity dilemma in continual learning via Kramers escape theory.
First comprehensive benchmark for evaluating AI models on professional graphic design tasks including layout, typography, and design intent translation.
ML-driven workflow for multi-objective discovery in materials science using automated microscopy and characterization, avoiding premature convergence.
Study analyzing bias toward American English in LLMs through postcolonial lens, examining how data curation and geopolitical histories shape model development.
Asymptotic convergence analysis of Q-learning with linear decay to zero learning rates addressing persistent bias and slow convergence issues.
Formal framework and metrics for pedagogical safety in educational reinforcement learning, introducing Reward Hacking Severity Index to detect misalignment.
Extension of transmission neural network model incorporating inhibitory connections and neurotransmitter populations with firing probability characterization.
Combee framework for scaling prompt learning in LLM agents enabling efficient self-improvement through system prompt optimization across parallel runs.
MC-CPO method for constrained reinforcement learning in tutoring systems preventing reward hacking through mastery-conditioned safety constraints.
Theoretical work on Expectation Propagation convergence avoiding non-integrable beliefs and optimizing constrained Bethe Free Energy.
Position paper analyzing failure modes in agentic IR systems where early errors cascade despite linguistic fluency, causing misalignment between reasoning and execution.
Open foundation models for Radio Access Network time-series forecasting enabling AI-native optimization and closed-loop control with improved generalization.
System and analysis of personalized LLM customization for individual investor decision-making, identifying fundamental limitations in current personalization paradigms.
Out-of-air computation framework for structured extraction from wireless superposition using joint source-channel coding without pre-embedded computation.
Neural Willmore flow approach using neural architectures and PINN-style loss functions to minimize Willmore energy on 2D surfaces.
Soft Tournament Equilibrium framework for evaluating LLM-based agents in non-transitive competitive settings using set-valued rankings instead of linear orderings.
Thermodynamics-inspired GeoAI method for modeling spatial heterogeneity and critical transitions in geographic and environmental systems.
Analysis of adversarial robustness in cloud-assisted autonomous driving systems with deep learning perception models and network vulnerabilities.
REAM method for pruning mixture-of-experts in large language models by merging experts, addressing memory challenges in deployment of billion-parameter models.
Theoretical analysis of integer-only operations for extreme learning machine classifiers to reduce computational cost at test time without accuracy loss.
Proposes methods to improve LLM agent performance at test-time without parameter updates by optimizing inference-time computation for complex reasoning tasks.
Identifies sparse routing mechanisms in alignment-trained LLMs using gate and amplifier heads to control refusal behavior, validated across 9 models from 6 labs.
Framework analyzing how ambient AI systems through causal user coupling transition from modeling to constituting part of cognitive function.
Model-based reinforcement learning approach for vehicle braking controller calibration.
Research on depth map super-resolution using semantic priors and RGB guidance.
Research on framework-agnostic quantum machine learning neural networks to reduce vendor lock-in across QML platforms.
Research on autonomous agents using multi-agent reinforcement learning for explainable cyber defense against APT techniques.
Generative model using conditional flow matching to simulate granular flow physics on inclined planes for inverse reconstruction.
Research on measuring consistency of model explanations across similar inputs using attribution stability metrics for explainable AI systems.
Defense framework against backdoor attacks in multimodal LLMs using patch-based cross-view regularization during fine-tuning.
Self-supervised contrastive learning framework for recommendation systems fusing long-term and short-term user interest patterns.
Optimization framework for automated water gate control balancing safety and performance requirements for Danish fjord management.
FLOWGEM: Iterative generative method using Wasserstein gradient flows for data imputation with non-monotone missingness patterns.
Adaptive learning approach for trajectory prediction in autonomous vehicles addressing long-tail distribution of safety-critical scenarios.
Statistical method for handling non-reciprocal pairwise comparisons in decision analysis and preference modeling with noise calibration.
Optimization technique integrating layout propagation into GEMM operations to reduce memory overhead in sequential matrix multiplications for ML workloads.
Kolmogorov-Arnold Networks applied to interpret crystalline energy landscapes for physics-informed property prediction with improved explainability.
Zero-shot depth reconstruction from UAV imagery using diffusion models for real-time geospatial tasks without task-specific retraining.
Studies optimality of Bayesian neural networks through statistical decision theory lens, analyzing minimaxity and admissibility properties.
Policy-driven model reconstructs protein residue networks to predict folding pathways correlating with experimental folding rates.
Fine-tuning integrity verification for neural networks using norm, rank, and sparsity certificates to detect backdoors and unauthorized changes.
Protocol enabling two AI agents to conduct secret conversations while producing transcripts indistinguishable from normal interaction.
Domain-specific BERT model trained on Turkish legal texts for NLP applications in legal technology.
Multi-modal sensor fusion framework using hybrid attention for 3D object detection in autonomous driving.
Compressed sensing with hybrid deterministic-random sampling from unitary matrix rows, providing denoising guarantees.
SkillX framework automatically constructs reusable skill knowledge bases for LLM agents, enabling efficient learning and generalization across tasks.
Hybrid quantum-classical Fourier Neural Operator for surrogate modeling of laser processing in PDE solvers.
Sparse identification of nonlinear dynamics with autoencoder for discovering system equations from noisy data.
Synthetic sandbox environment for training ML engineering agents that can handle expensive ML verification tasks via fast mock pipelines.