PEANUT proposes perturbation-based adversarial attack method targeting robustness vulnerabilities in Graph Neural Networks through topology modifications.
PruneFuse strategy uses pruned networks for efficient data selection and fuses them with original networks to optimize deep neural network training.
Complexity analysis of optimal graph rewiring to address oversmoothing and oversquashing in deep graph neural networks.
Unified framework for data-centric dynamic training of LLMs with consistent interfaces for data selection and reweighting.
Study of LLM-based AI scientist agents learning from iterative experimental feedback in cell screening with 800 replicated experiments.
Graph neural network framework using Ricci flow for improved geometric representation learning on graphs.
Analysis of optimization trade-offs in asynchronous federated learning addressing gradient staleness and client bias.
Knowledge distillation approach for deploying Transformer-based reinforcement learning on resource-constrained energy management devices.
Machine learning risk score for hypertrophic cardiomyopathy combining echocardiography and clinical data.
Conformal prediction extension for contrastive learning with learnable multi-norm constraints and coverage guarantees.
Topology-aware reinforcement learning with GNNs for optimal dispatch of energy storage systems in distribution networks.
Graph attention network for ADHD identification using dynamic functional connectivity from fMRI data.
Expected free energy acquisition function for Bayesian optimization with convergence guarantees and connections to existing methods.
Formal framework for measuring uncertainty in LLM text generation accounting for prompting, generation, and interpretation stages.
Survey of generative modeling in protein design covering neural representations, conditional generation, and evaluation standards.
Margin scheduler for triplet loss in Siamese networks to improve distance metric learning training dynamics.
Conformal prediction method for uncertainty quantification under covariate shift using selective kernel mean matching.
Theory-informed deep learning model for long-term traffic forecasting balancing interpretability and predictive accuracy.
Neuro-symbolic approach combining neural networks and domain knowledge for process anomaly detection from event logs.
Koopman autoencoder-based least-squares policy iteration algorithm enabling automatic feature learning in reinforcement learning.
Transformer enhanced with Boltzmann machines for DNA sequence classification to uncover latent interactions and higher-order dependencies.
Framework using Shapley values to measure and explain unfairness in machine learning models under group fairness criteria with inference methods.
SPECTRA: spectral-informed neural network for sensor-based activity recognition optimized for edge deployment with low latency and privacy.
EcoFair: privacy-preserving medical inference framework with lightweight routing for vertically partitioned data and modality-specific embeddings.
Theoretical analysis of spectral optimizers like Muon in language model training, studying capacity scaling through linear associative memory framework.
Machine unlearning framework addressing retain-forget entanglement where retained samples unintentionally affected by forgetting correlated features.
Deep learning models for fear and muscle activity correlation in climbing sports using psychophysiological data. Niche domain application.
Study comparing sample selection methods (random, farthest-first, interactive visualization) for annotation of biomedical time-series data with real annotators.
PQuantML: open-source hardware-aware neural network compression library for pruning and quantization with unified interface for latency-constrained deployment.
Machine learning approach to forecasting solar power ramp events for grid stability. Energy domain application.
Quantum-inspired anomaly detection using hardware-aware tensor networks for particle collider physics, deployable on classical hardware.
Systematic evaluation of tabular foundation models like TabPFN and TabICL for conditional density estimation in regression tasks with heteroscedasticity.
C²MF: context-specific credibility-aware multimodal fusion framework using probabilistic circuits to handle conflicting modalities and situational reliability changes.
Framework for marginalizing latent parameters in Bayesian models via automatic differentiation combined with nested sampling.
LP-based sampling policy for multi-armed bandits with network-enabled side-observations and stochastic action availability.
Lightweight self-adaptive ML framework for DC arc-fault detection in photovoltaic systems addressing hardware heterogeneity and environmental noise.
Kolmogorov-Arnold Network ensemble learning for early hit enrichment in virtual screening with improved positive predictive value metrics.
Method for automatic timestep selection in Diffusion Transformers to improve efficiency of discriminative representation learning.
Variational autoencoder with adaptive Hidden Markov priors for unsupervised blind source separation with source-specific temporal modeling.
Judge Agent system using automated mathematical validation to reduce silent failures in LLM-generated scientific simulation code from 42% to 1.5%.
Vision Transformers and GNNs for charged particle track reconstruction in ATLAS muon spectrometer under high luminosity conditions.
Theoretical guarantees for learning causal representations from limited environments and finite samples.
Study evaluating whether Vision Transformer architectures universally require register tokens to eliminate attention map artifacts.
LLM framework for formal proof repair using counterexample-guided reasoning and behavioral feedback to improve automated verification.
Comprehensive evaluation framework for agent-based medical AI systems via multi-step clinical dialogue simulation with realistic physician-patient interactions.
Neural score-based particle method for simulating collisional plasma kinetics in the Vlasov-Maxwell-Landau system.
Dataset and VLM method for generating whiteboard animations synchronized with speech narration across STEM domains using structured drawing representations.
Real-world evaluation of visual navigation foundation models on robot navigation, testing generalization and providing trajectory quality metrics.
Vision-language learning approach for end-to-end autonomous driving using multimodal datasets and collision-aware representation learning.
Optimization framework for robust decisions when predictions lack calibrated error bounds, combining robust and regret formulations.