The Archive

Eye-tracking metrics (pupil diameter, fixation) classify left/right brain hemisphere cognitive activity.

Parametric decomposition method for analyzing sub-breath respiratory airflow signals in medical contexts.

Associativity-peakiness metric for clustering algorithm contingency table evaluation.

FedSPDnet introduces geometry-preserving federated learning aggregation strategies for symmetric positive definite matrices with Stiefel constraints.

LLMs evaluated as semantic evaluators for ASR, achieving 92-94% human agreement compared to traditional WER metrics.

Computational framework analyzes Brazilian parliamentary discourse using stylometric analysis and topic modeling on legislative speech.

Bayesian Optimal Experimental Design framework using integral probability metrics replaces KL divergence for stable information gain estimation.

Quotient-Space Diffusion Models framework formalizes generative modeling with symmetries, applied to 3D molecular structure generation.

GEM: smooth rational activation functions matching ReLU performance with C^2N differentiability for deep networks.

SAIL: solver-aligned initialization learning improves SCF convergence for molecular geometry by optimizing supervision targets, not extrapolation.

Geometric characterization of trajectory matching for clinical dataset condensation reveals supervision signal structure and synthetic data scaling.

Multivariate Kernel Score for conformal prediction adapts to residual geometry and connects Bayesian to frequentist uncertainty quantification.

Gauge-equivariant graph neural networks embed non-Abelian symmetries into message passing for lattice gauge theory problems.

SWE-chat dataset of 6,000 real coding agent sessions from open-source developers provides empirical usage patterns and output utility metrics.

MGDA-Decoupled balances multiple alignment objectives in DPO-based LLM training via geometry-aware optimization.

Multi-scale metric on strings using n-gram angle distances with exponential weights, proven metric properties and linear-time algorithm.

Occupancy Reward Shaping uses optimal transport on world models to extract temporal geometry for credit assignment in offline goal-conditioned RL.

Multilingual medical QA evaluation shows cross-lingual evidence retrieval and LLM parametric knowledge improve performance on low-resource languages relative to monolingual approaches.

Knowledge Capsules add structured nonparametric memory units to LLMs for stable multi-hop reasoning and knowledge updates without retraining.

Theoretical analysis connecting random decision trees to overlap gap properties in symmetric binary perceptrons; advances statistical computational gap characterization.

Reproducibility study standardizing evaluation of 11 counterfactual explanation methods for recommender systems across metrics and datasets.

Inference-time hallucination detection in SpeechLLMs via attention metrics (AUDIORATIO, AUDIOCONSISTENCY); avoids gold-standard requirement.

Theoretical framework decoupling geometry from probability assumptions in classical generalization analysis.

Multi-agent LLM framework addresses actor-observer cognitive bias via dialectical alignment in self-reflection loops.

Dual-track LLM-assisted stylometry (embedding + register) detects biblical allusions in Cormac McCarthy via KJV vocabulary and semantic matching.

DASH-KV accelerates long-context LLM inference via asymmetric KV cache hashing to reduce attention complexity.

EDGE-EVAL framework assesses LLM deployment viability via energy, latency, cost constraints beyond accuracy metrics.

RDP LoRA uses geometry-driven trajectory analysis to identify optimal LoRA adapter placement in LLM fine-tuning without training.

DOPE: semiparametric estimator for scalar functionals from neural operator solution trajectories with debiasing.

Unsupervised defect detection integrates Denoising Diffusion Probabilistic Model with asymmetric teacher-student network for industrial surface inspection.