Results 181 to 190 of about 3,107 (258)

Sampling Strategy: An Overlooked Factor Affecting Artificial Intelligence Prediction Accuracy of Peptides’ Physicochemical Properties

open access: yesAdvanced Intelligent Discovery, EarlyView.
This study reveals that sampling strategy (i.e., sampling size and approach) is a foundational prerequisite for building accurate and generalizable AI models in peptide discovery. Reaching a threshold of 7.5% of the total tetrapeptide sequence space was essential to ensure reliable predictions.
Meiru Yan   +3 more
wiley   +1 more source

Interpretability and Representability of Commutative Algebra, Algebraic Topology, and Topological Spectral Theory for Real‐World Data

open access: yesAdvanced Intelligent Discovery, EarlyView.
This article investigates how persistent homology, persistent Laplacians, and persistent commutative algebra reveal complementary geometric, topological, and algebraic invariants or signatures of real‐world data. By analyzing shapes, synthetic complexes, fullerenes, and biomolecules, the article shows how these mathematical frameworks enhance ...
Yiming Ren, Guo‐Wei Wei
wiley   +1 more source

AI Powered Biobanks From Static Archives to Dynamic Discovery Engines

open access: yesAdvanced Intelligent Discovery, EarlyView.
Large language models (LLMs) provide a potential framework for transforming biobanks from static data repositories into intelligent discovery engines. By enabling unified representation and analysis of multimodal biomedical data, LLM‐based systems facilitate dynamic risk prediction, biomarker identification, and mechanistic interpretation, thereby ...
Wenzhen Yin   +5 more
wiley   +1 more source

The Interoperability Challenge in DFT Workflows Across Implementations

open access: yesAdvanced Intelligent Discovery, EarlyView.
Interoperability and cross‐validation remain major challenges in the computational materials science. In this work, we introduce a common input/output standard that enables internal translation across multiple workflow managers—AiiDA, PerQueue, Pipeline Pilot, and SimStack—while producing results in a unified schema.
Simon K. Steensen   +13 more
wiley   +1 more source

Data‐Efficient Cycle‐Level Capacity Prediction Using 1D Deep Convolutional Network

open access: yesAdvanced Intelligent Discovery, EarlyView.
We introduce DeepBat, a deep learning framework featuring a 1D convolutional backbone designed to extract latent degradation patterns from a microstructurally diverse electrode dataset. By learning complex formulation–performance relationships, the model accurately predicts long‐term specific discharge capacity using limited early‐cycle data, providing
Tao Huang   +16 more
wiley   +1 more source

Home - About - Disclaimer - Privacy