A foundation model for the Earth system
A single AI model was applied to predict diverse Earth system phenomena such as weather, waves, air pollution, and typhoon paths.
Published 2025.05.21View evidence and readiness →Artificial Intelligence
Explore artificial intelligence methods and their applications in science and industry.
Mastering diverse control tasks through world models
The study demonstrates that a single AI system can learn to control various tasks by predicting environmental outcomes using world models.
Published 2025.04.02View evidence and readiness →A generative model for inorganic materials design
A generative model called MatterGen was developed to create stable inorganic materials with specific properties.
Published 2025.01.16View evidence and readiness →Mechanical performance of steel, basalt and polyamide fiber-reinforced ultra-high-performance concrete evaluated by experiments and SHAP-based machine learning
This study evaluates the mechanical performance of fiber-reinforced ultra-high-performance concrete using experiments and SHAP-based machine learning to analyze non-linear material behavior.
Published 2026-10-07View evidence and readiness →A hybrid machine learning framework integrating persian NLP for occupational risk classification in steel manufacturing
This study presents a hybrid machine learning framework that combines structured risk data with Persian natural language processing to improve hazard classification accuracy in steel manufacturing environments.
Published 2026-10-02View evidence and readiness →Deep reinforcement learning-based adaptive FOPID tuning for power system stability enhancement: A twin delayed deep deterministic policy gradient approach
This paper presents a TD3-based adaptive FOPID controller that automatically adjusts parameters to improve power system stability with renewable energy.
Published 2026-10-02View evidence and readiness →Paradigm shift from statistical channel modeling to digital twin prediction: an environment-generalizable ChannelLM for 6G AI-enabled air interface
This paper proposes a channel large model (ChannelLM) that uses AI to predict wireless channels in real-time across various environments, offering better generalization than traditional statistical models.
Published 2026-10-01View evidence and readiness →From Static to Dynamic: Benchmarking Real-World Code Review with MCR-Bench
This study introduces MCR-bench, a benchmark for realistic multi-round code review that includes detailed defect information and state labels across multiple review rounds.
Published 2026-10-01View evidence and readiness →A Centralized V2I LoRaWAN Architecture for Real-Time Adaptive Route Planning and Traceability of Electric Vehicles
This paper proposes a centralized V2I LoRaWAN architecture that enables real-time adaptive route planning and traceability for electric vehicles using AI-based algorithms and LoRaWAN communication.
Published 2026-10-01View evidence and readiness →Multimodal AI for Documents, Images, Audio, and Video: Production Architecture, Current Research, Real-World Use Case, Implementation, and Verification
This paper presents a comprehensive production-ready AI architecture for handling documents, images, audio, and video data in real-world applications.
Published 2026-09-29View evidence and readiness →World Models for Simulation, Planning, and Agent Training: Production Architecture, Current Research, Real-World Use Case, Implementation, and Verification
This paper presents a comprehensive AI architecture for training agents in simulated environments, with real-world application in robotics.
Published 2026-09-29View evidence and readiness →Experimental validation of irradiance-sensorless self-adaptive GMPPT algorithm for PV systems under partial shading
This study experimentally validates an irradiance-sensorless self-adaptive global MPPT algorithm for PV systems under partial shading conditions, demonstrating its ability to identify the true maximum power point without external irradiance measurements.
Published 2026-09-29View evidence and readiness →Energy-efficient clustering and routing protocol for wireless sensor networks based on improved HOA and attention-guided D3QN
This study proposes HOADRL, an energy-efficient clustering and routing protocol for wireless sensor networks that uses an improved hiking optimization algorithm (HOA) and deep reinforcement learning to optimize cluster formation and reduce energy consumption.
Published 2026-09-28View evidence and readiness →Graph Neural Network‐Based Reinforcement Learning for Decentralized Multi‐Robot Manipulation
This study presents a graph neural network-based framework for decentralized multi-robot manipulation, enabling effective communication and coordination among agents without centralized planning.
Published 2026-09-26View evidence and readiness →Bhygnn+: unsupervised representation learning for heterophilic hypergraphs
BHyGNN+ is a self-supervised learning framework that extends BHyGNN for representation learning on heterophilic hypergraphs without requiring ground-truth labels.
Published 2026-09-26View evidence and readiness →Performant Tridiagonal Factorization of Skew-Symmetric Matrices
This paper introduces an efficient algorithm for factorizing skew-symmetric matrices into a unit lower-triangular matrix and a tridiagonal matrix, enabling faster computation of determinants and solving systems of equations.
Published 2026-09-25View evidence and readiness →DR-GEM enables self-supervised machine learning for single-cell embeddings and annotations
DR-GEM is a self-supervised machine learning method that improves single-cell data analysis by focusing on rare cell types and patterns through reconstruction error and balanced consensus learning.
Published 2026-09-24View evidence and readiness →Mixture replication in benchmark datasets inflates reported machine learning performance and degrades prediction interval coverage for ultra-high-performance concrete
The study demonstrates that the predictive performance of machine learning models for ultra-high-performance concrete is inflated by the presence of replicate records in benchmark datasets, and that mixture-grouped cross-validation provides a more accurate evaluation method.
Published 2026-09-24View evidence and readiness →Squirrel OS: A Self-Governing Neural Mesh Operating System with Deterministic Healing, Post-Quantum Cryptographic Remediation, and Recursive Intelligence — Industry Technical Paper (v1.3 Public-Release Edition)
This industry technical paper presents the Squirrel OS self-governing neural mesh operating system with deterministic healing, post-quantum cryptographic remediation, and recursive intelligence, along with architecture, benchmarks, and ecosystem results.
Published 2026-09-20View evidence and readiness →Optimizing energy consumption in industrial wireless sensor networks using a hybrid FCM and GWO machine learning approach
This study proposes a hybrid machine learning algorithm combining Fuzzy C-means and Grey Wolf Optimization to optimize energy consumption in industrial wireless sensor networks.
Published 2026-09-18View evidence and readiness →Quantum Algorithms for Trading: A Survey of Speedups, Thresholds, and Dequantization
This paper reviews quantum algorithms for trading, highlighting their potential speedups through amplitude estimation and the challenges in achieving practical runtime advantages over classical methods.
Published 2026-09-17View evidence and readiness →Machine-Learned Mismatch and Task Preservation Beliefs in CoSMA DAI for Common Knowledge Aware Semantic Alignment
CoSMA DAI uses causal machine learning to detect and repair semantic mismatches in communication systems while preserving task integrity through transmitter-derived beliefs and BDIx agents.
Published 2026-09-17View evidence and readiness →Dynamic voltage stability enhancement in wind integrated power systems using a deep reinforcement learning based FACTS controller
This study proposes a deep reinforcement learning (DRL) based control system for FACTS devices to enhance reactive power management and dynamic voltage stability in wind-integrated power systems.
Published 2026-09-16View evidence and readiness →Fault tolerant cluster-based routing in wireless sensor network using novel intelligent deep learning methodology
This study proposes a fault-tolerant cluster-based routing model that combines an Improved Graph Neural Network with Multi-Objective Crested Porcupines Optimization to recover from cluster head failures in wireless sensor networks.
Published 2026-09-15View evidence and readiness →A multistage CT-based artificial intelligence system for microvessel density quantification in lung adenocarcinoma: correlation with histopathology
A multistage AI system was developed to quantify microvessel density in lung adenocarcinoma using CT images, showing correlation with histopathological results.
Published 2026-09-14View evidence and readiness →Research on assessment and adaptive correction technology of power system protection fixed value mismatch
This study presents an adaptive current protection setting optimization method for power systems that considers multiple objectives and constraints to improve reliability and selectivity.
Published 2026-09-13View evidence and readiness →Data-driven and experimental optimization of thermal performance in ZrO₂ Nanofluid oscillating heat pipes
This study optimizes the thermal performance of ZrO₂–acetone nanofluid oscillating heat pipes using Response Surface Methodology and Machine Learning techniques.
Published 2026-09-13View evidence and readiness →Data-driven cycle time optimisation in semiconductor manufacturing using machine learning and heuristic search algorithms
This paper presents a hybrid framework combining machine learning and heuristic optimization to reduce cycle time in semiconductor manufacturing by predicting and optimizing key performance indicators.
Published 2026-09-12View evidence and readiness →Distributed learning-enabled federated large language model for fake news detection
This research proposes a biometric authenticated-distributed learning enabled federated large language model (Bio-DFLLM) to improve fake news detection by enhancing privacy and reducing computational complexity.
Published 2026-09-12View evidence and readiness →Organizational strategic decision support model in textile manufacturing enterprises under deep learning and knowledge graph
This study develops an integrated decision-making framework for textile manufacturing using knowledge graphs, deep learning, and reinforcement learning to address data fragmentation and improve production efficiency.
Published 2026-09-11View evidence and readiness →Deep Reinforcement Learning with State Uncertainty Estimation for Continuous Control of Dynamic Complex Systems
This paper proposes a deep reinforcement learning framework that integrates state uncertainty estimation with policy learning to improve control in dynamic complex systems.
Published 2026-09-10View evidence and readiness →External validation of a deep learning model for hip fracture detection on pelvic radiographs and its effect on radiology resident performance
This study externally validated a deep learning model for hip fracture detection on pelvic radiographs and found that AI assistance improved radiology resident performance.
Published 2026-09-10View evidence and readiness →Reducing cost of fluorescent microscopy through dye exclusion, distillation techniques, and deep learning
This study presents CR-DINO, a self-distillation method that reduces the number of fluorescent channels needed for high content screening while maintaining image quality.
Published 2026-09-09View evidence and readiness →A robust XAI guided structured pruning approach for low-complexity and high-performance food classification
This study proposes a structured pruning approach for food classification using a Grad-CAM inspired channel importance criterion to enhance model efficiency while maintaining performance and interpretability.
Published 2026-09-08View evidence and readiness →Integrating waste heat and biomass energy in a three-entropy state framework optimized by emergy-carbon-neural network for sustainable buildings
This study proposes a novel building energy system integrating industrial waste heat and biomass energy, optimized by an emergy-carbon-neural network model, to improve energy efficiency and reduce carbon emissions.
Published 2026-09-07View evidence and readiness →An Integrated AI Automating Inspection Model for Surface Defect Segmentation and Classification in Construction Domain
This study presents an AI-based automated inspection model for detecting and classifying surface defects in construction environments, aiming to improve efficiency and safety compared to manual methods.
Published 2026-09-06View evidence and readiness →ConvSNN: an energy-efficient hybrid convolutional-spiking neural network for wrist-worn stress and affective state recognition
This study presents ConvSNN, a hybrid convolutional-spiking neural network designed for energy-efficient stress and affective state recognition using wrist-worn physiological signals.
Published 2026-09-06View evidence and readiness →A hybrid memetic algorithm for an energy-aware distributed blocking hybrid flowshop scheduling problem with transportation time
This paper proposes a hybrid memetic algorithm with three-layer encoding and adaptive local search to minimize makespan and energy consumption in distributed blocking hybrid flowshop scheduling with transportation time.
Published 2026-09-06View evidence and readiness →Novel adaptive fixed-time synchronization for fractional-order uncertain Hopfield neural networks with proportional delays of unbounded nature
This paper proposes an adaptive controller for fixed-time synchronization of fractional-order uncertain Hopfield neural networks with unbounded proportional delays, using multiple modules to enhance convergence and suppress delay effects.
Published 2026-09-06View evidence and readiness →A multi-task deep learning network for robust and high accuracy star sensor image centroiding
This study presents a lightweight multi-task CNN that simultaneously classifies single and binary stars while estimating sub-pixel centroids with high accuracy and minimal computational overhead.
Published 2026-09-06View evidence and readiness →A study on the GraphRAG semantic retrieval algorithm for multimodal data
This paper presents a GraphRAG model that integrates text, image, and structured data into a unified semantic space for improved retrieval performance.
Published 2026-09-05View evidence and readiness →ConvLSTMNet-EMD: leveraging 1D-convolutional encoder-decoder model with self-attention mechanism for improved energy consumption forecasting
This paper presents ConvLSTMNet-EMD, a framework combining 1D-convolutional encoder-decoder with LSTM and EMD to improve peak demand forecasting in short-term load prediction.
Published 2026-09-05View evidence and readiness →Novel analysis of physics-informed neural networks for solving Cahn-Allen, FitzHugh-Nagumo and Fisher–KPP Models
This study introduces a hybrid method combining the New Iterative Method with Physics-Informed Neural Networks to solve nonlinear partial differential equations more efficiently.
Published 2026-09-05View evidence and readiness →Predicting survival of patients with deep burns using supervised machine learning algorithms
A machine learning model was developed to predict in-hospital mortality in deep burn patients using early admission data and validated with cross-validation techniques.
Published 2026-09-04View evidence and readiness →Prediction of dielectric behaviour of FDM printed-acrylonitrile butadiene styrene (ABS) reinforced with 10% carbon fiber: a machine learning approach
This study uses machine learning to predict dielectric properties of 3D-printed acrylonitrile butadiene styrene (ABS) reinforced with 10% carbon fiber based on printing parameters.
Published 2026-09-03View evidence and readiness →A novel deep learning-based fraud detection framework integrating autonomous PSO-GA-based feature selection, hyperparameter optimization, and explainable artificial intelligence
This paper presents a hybrid deep learning framework for fraud detection that integrates autonomous PSO-GA-based feature selection, hyperparameter optimization, and explainable AI to improve accuracy and interpretability.
Published 2026-09-03View evidence and readiness →CoreMind — An Alternative Architecture for Explainable, Efficient, and Evolving Artificial Intelligence
CoreMind is an AI architecture that uses structured tensor representations, matrix-based semantics, and custom attention mechanisms to enable explainable, efficient, and evolving artificial intelligence.
Published 2026-09-02View evidence and readiness →Designing physics experiments with artificial intelligence
This study explores the use of artificial intelligence in designing physics experiments by searching for optimal hardware configurations and proposing new experimental layouts.
Published 2026-09-02View evidence and readiness →Multi agent PPO for terahertz cell free mobile edge computing networks
This study proposes a multi-agent PPO algorithm for terahertz cell-free mobile edge computing networks to optimize energy consumption and latency.
Published 2026-08-31View evidence and readiness →Transformer Fault Classification Based on Dissolved Gas Compositions Using Microsoft Azure Machine Learning
This study develops a machine learning model using Microsoft Azure to classify transformer faults based on dissolved gas analysis without requiring programming skills.
Published 2026-08-30View evidence and readiness →Application of artificial intelligence-powered CNN in image recognition and inheritance of southwest traditional crafts
This study proposes a new AI model called SW-CraftNet that uses multi-scale features and attention mechanisms to improve the recognition of southwest traditional crafts.
Published 2026-08-30View evidence and readiness →Digital transformation capabilities and innovation performance in skating manufacturing: a hybrid SEM–ANN analysis
This study uses a hybrid SEM-ANN model to analyze how digital transformation capabilities affect innovation performance in skating equipment manufacturing firms in China.
Published 2026-08-29View evidence and readiness →Neural network–augmented Pfaffian wave-functions for scalable simulations of interacting fermions
This study proposes a neural network–augmented Pfaffian wave-function called the hidden fermion Pfaffian state (HFPS) that effectively represents both unpaired and superconducting phases in fermionic systems with scalable computational complexity.
Published 2026-08-28View evidence and readiness →Machine learning-guided discovery of natural anti-inflammatory agents targeting NF-κB p65
This study presents a machine learning framework to identify natural anti-inflammatory agents targeting NF-κB p65, a key protein in inflammatory signaling pathways.
Published 2026-08-28View evidence and readiness →Large language models as uncertainty-calibrated optimizers for experimental discovery
This study demonstrates that training large language models with Bayesian objectives allows them to function as reliable optimizers in experimental discovery by incorporating uncertainty estimates.
Published 2026-08-28View evidence and readiness →Performance of multi-strategy optimized mayfly optimization algorithm for location selection of university reimbursement form submission terminals
This study proposes a multi-strategy optimized mayfly optimization algorithm (MSMOA) to improve the performance of the Mayfly Optimization Algorithm (MOA) for location selection of university reimbursement form submission terminals.
Published 2026-08-27View evidence and readiness →Premixed cocktail configurations for media blending altered cell culture outcomes discovered by active learning
Machine learning was used to identify optimal premixed ingredient configurations that significantly enhance cell culture outcomes compared to standard media.
Published 2026-08-27View evidence and readiness →Agentic Data Engineering Framework for Autonomous, Real-Time, and Trustworthy AI Systems
This paper proposes an Agentic Data Engineering Framework that uses a multi-agent system to dynamically reconfigure data pipelines for autonomous, real-time, and trust-conscious data processing.
Published 2026-08-25View evidence and readiness →Research on multi-information sensing and graded warning technology for fires in electromechanical chambers
This research develops a multi-sensor system with deep learning for real-time fire monitoring and early warning in underground coal mine electromechanical chambers.
Published 2026-08-25View evidence and readiness →Missing physics discovery through fully differentiable finite element-based machine learning
This study presents FEML, a differentiable framework that couples known partial differential equations with machine-learned operators to discover missing physical laws from PDE solutions.
Published 2026-08-25View evidence and readiness →Multi-resolution enhancement for full-spectrum neural representations
This paper presents a hierarchical neural representation framework that enhances the ability of smaller networks to capture detailed and multiscale structures in scientific data.
Published 2026-08-24View evidence and readiness →Machine learning enhanced prediction of optical properties via reverse Monte Carlo method
A machine learning approach is used to accelerate the extraction of optical properties from REELS spectra using the reverse Monte Carlo method.
Published 2026-08-24View evidence and readiness →Quantitative Assessment of LiDAR Availability in Smoke-Filled Tunnels Using a Degradation Scoring Algorithm
This paper presents a degradation scoring algorithm to quantitatively assess LiDAR performance in smoke-filled tunnels using physical indicators and a visibility reference.
Published 2026-08-24View evidence and readiness →Machine learning of artistic fingerprints in jazz
Cheston et al. developed a machine learning pipeline to identify 20 iconic jazz pianists from audio recordings with high accuracy, showing how musical elements shape individual performance styles.
Published 2026-08-17View evidence and readiness →