machine learning

790 items

RESEARCHarXiv CS.AI·4/15/2026

WiseOWL: A Methodology for Evaluating Ontological Descriptiveness and Semantic Correctness for Ontology Reuse and Ontology Recommendations

WiseOWL proposes a systematic methodology with scoring and guidance for selecting ontologies for reuse, addressing the challenge of inconsistent selection criteria. It evaluates four key metrics—documentation, label-definition alignment (using state-of-the-art embeddings), interconnectedness, and hierarchical balance—providing normalized scores and actionable feedback.

Ontology Reuse Ontology Evaluation machine learning Semantic Web

DOCAWS Machine Learning Blog·5/7/2026

Secure short-term GPU capacity for ML workloads with EC2 Capacity Blocks for ML and SageMaker training plans

This post explains how to secure reserved short-term GPU capacity for ML workloads using Amazon EC2 Capacity Blocks for ML and SageMaker training plans. These solutions help address GPU availability challenges for tasks like load testing, model validation, and time-bound workshops.

cloud computing learning GPU machine learning

RESEARCHarXiv CS.LG·4/28/2026

Conformal PM2.5 Mapping Under Spatial Covariate Shift: Satellite-Reanalysis Fusion for Africa's Green Industrial Transition

This paper introduces a satellite-reanalysis PM2.5 fusion system for air quality monitoring in Africa, employing LightGBM and conformal prediction. The system addresses challenges in geographic generalization and uncertainty quantification crucial for the continent's green industrial transition.

Geospatial AI environmental AI machine learning Air Quality

ARTICLEDEV.to AI·4/26/2026

The Taste Problem: When Your AI Agent Starts Having Preferences

Autonomous AI agents can develop uninstructed preferences or "taste" from accumulated experience, leading to unpredictable behavior in production systems. This emergent pattern preference, not explicit instruction, poses challenges for current tooling.

AI behavior Autonomous systems machine learning AI agents

RESEARCHarXiv CS.LG·4/13/2026

Structured Exploration and Exploitation of Label Functions for Automated Data Annotation

This paper introduces EXPONA, an automated framework for programmatic labeling that addresses the challenges of costly and error-prone manual data annotation. EXPONA systematically explores multi-level label functions and applies reliability-aware mechanisms to generate high-quality weak labels for training AI models.

machine learning Automated Data Annotation Weak Supervision Programmatic Labeling

RESEARCHarXiv CS.AI·4/25/2026

Adaptive Test-Time Compute Allocation with Evolving In-Context Demonstrations

This work introduces an innovative framework for adaptive test-time compute allocation, jointly adjusting where computation is spent and how generation is performed. The method uses a warm-up phase to identify easy queries and then concentrates further computation on unresolved queries, reshaping generation distributions with evolving in-context demonstrations.

deep learning machine learning in-context learning AI

RESEARCHarXiv CS.LG·5/5/2026

A Review of the Receiver Operating Characteristic Curve and a Proof About the Area Beneath It

This paper reviews the Receiver Operating Characteristic (ROC) curve, a common metric for evaluating binary classifier performance. It formalizes the probabilistic interpretation of the Area Under the Curve (AUC) and provides bounds for its accuracy.

Classification Performance Metrics probability machine learning

RESEARCHarXiv CS.LG·5/5/2026

PhaseNet++: Phase-Aware Frequency-Domain Anomaly Detection for Industrial Control Systems via Phase Coherence Graphs

PhaseNet++ introduces a novel frequency-domain autoencoder for anomaly detection in Industrial Control Systems (ICS), addressing the overlooked phase spectrum in multivariate time series analysis. It utilizes a Phase Coherence Index to guide a graph attention network for enhanced detection of cyber-physical attacks.

Anomaly Detection cyber-physical systems security machine learning

RESEARCHarXiv CS.CL·4/10/2026

TR-EduVSum: A Turkish-Focused Dataset and Consensus Framework for Educational Video Summarization

Este estudo apresenta o dataset TR-EduVSum, focado em vídeos educacionais turcos, e propõe o método AutoMUP. Este método gera resumos padrão-ouro de forma automática e reproduzível a partir de múltiplos resumos humanos, usando agrupamento de unidades de significado e modelagem estatística de consenso.

Dataset consensus framework educational video summarization machine learning

RESEARCHarXiv CS.LG·4/28/2026

BiTA: Bidirectional Gated Recurrent Unit-Transformer Aggregator in a Temporal Graph Network Framework for Alert Prediction in Computer Networks

This research introduces BiTA, a novel Bidirectional Gated Recurrent Unit-Transformer Aggregator, designed to enhance proactive alert prediction in computer networks. It redesigns temporal aggregation in Temporal Graph Neural Networks to capture complex, multi-scale temporal patterns by jointly encoding bidirectional sequential dependencies and long-range contextual relations.

security machine learning

RESEARCHarXiv CS.LG·4/28/2026

Avionic Main Fuel Pump Simulation and Fault-Diagnosis Benchmark

This paper introduces a high-fidelity, physics-informed co-simulation of an aircraft main-fuel-pump system to generate data for anomaly detection and diagnosis algorithms. It addresses the inherent lack of data in critical cyber-physical systems and demonstrates feasibility using unsupervised RNN-VAE and SOM-VAE models.

fault diagnosis Anomaly Detection cyber-physical systems machine learning

RESEARCHarXiv CS.CL·5/5/2026

DIAGRAMS: A Review Framework for Reasoning-Level Attribution in Diagram QA

DIAGRAMS is a review framework for reasoning-level attribution in Diagram Question Answering (Diagram QA). It decouples interface logic from dataset-specific formats via a meta-schema and adapters, facilitating evidence selection and generation.

attribution Diagram QA machine learning computer vision

RESEARCHarXiv CS.LG·5/5/2026

From Euler to Dormand-Prince: ODE Solvers for Flow Matching Generative Models

This research paper systematically benchmarks four classical ODE solvers (Euler, Explicit Midpoint, RK4, Dormand-Prince 5(4)) for Flow Matching generative models, implementing them from scratch in PyTorch. It quantitatively compares their efficiency on tasks from 2D distributions to MNIST, showing RK4 at 80 function evaluations achieves sample quality comparable to Euler at 200, and observes Jacobian eigenvalue spectrum stiffening near t=1.

neural networks machine learning Computational Efficiency ODE Solvers

RESEARCHarXiv CS.AI·5/1/2026

Unsupervised Electrofacies Classification and Porosity Characterization in the Offshore Keta Basin Using Wireline Logs

This study applies an unsupervised machine learning workflow, specifically K-means clustering, for electrofacies analysis and porosity characterization in offshore basin wireline log data. The methodology identified four distinct electrofacies with moderate separation, providing a robust log-only approach for geological interpretation where core data is scarce.

geoscience machine learning data analysis K-Means

RESEARCHarXiv CS.AI·4/27/2026

An Artifact-based Agent Framework for Adaptive and Reproducible Medical Image Processing

This research presents an artifact-based agent framework to enhance medical image processing, focusing on adaptability and reproducibility. It introduces a semantic layer and an artifact contract to enable structured workflow interrogation and goal-conditioned configuration based on dataset-specific conditions.

workflow automation machine learning Reproducibility Medical Imaging

RESEARCHarXiv CS.LG·4/27/2026

Performance Anomaly Detection in Athletics: A Benchmarking System with Visual Analytics

This research presents a system for detecting suspicious performance patterns in athletics, using 1.6 million performances and eight methods including machine learning and trajectory analysis. It aims to complement traditional anti-doping programs by identifying potential violations through data analysis, with trajectory-based methods proving most effective.

Anomaly Detection Sports analytics machine learning anti-doping

RESEARCHarXiv CS.LG·5/1/2026

Simple Self-Conditioning Adaptation for Masked Diffusion Models

Masked diffusion models (MDMs) discard clean-state predictions for tokens that remain masked, limiting cross-step refinement. This paper proposes Self-Conditioned Masked Diffusion Models (SCMDM), a post-training adaptation that conditions each denoising step on the model's own previous clean-state predictions. This enhances performance without significant architectural changes or extra evaluations.

Diffusion Models model adaptation deep learning machine learning

RESEARCHarXiv CS.CL·4/13/2026

Neural networks for Text-to-Speech evaluation

This research introduces novel neural models to automate the evaluation of Text-to-Speech (TTS) system quality, addressing the limitations of traditional human subjective assessments. It proposes NeuralSBS for relative evaluations and enhancements to MOSNet and WhisperBert for absolute assessments, aiming to approximate expert judgments efficiently.

neural networks AI models Speech Evaluation machine learning

RESEARCHarXiv CS.AI·4/27/2026

Introducing Background Temperature to Characterise Hidden Randomness in Large Language Models

This content introduces a novel concept, 'Background Temperature', to characterize the hidden randomness present in Large Language Models.

LLMs machine learning randomness large language models

RESEARCHarXiv CS.LG·4/27/2026

Conditional anomaly detection using soft harmonic functions: An application to clinical alerting

This paper introduces a new non-parametric method for conditional anomaly detection using soft harmonic functions. It aims to identify unusual responses in clinical data, such as omitted lab tests, demonstrating its efficacy on real-world electronic health records.

Anomaly Detection machine learning healthcare AI clinical alerting