Results 121 to 130 of about 36,835 (256)

DDSurfer: A Weakly‐Supervised Dual‐Stream Deep Learning Framework for Cortical Surface Reconstruction From Diffusion MRI

open access: yesAdvanced Science, EarlyView.
DDSurfer reconstructs cortical surfaces directly from diffusion MRI without requiring T1‐weighted scans. By fusing complementary microstructural features and learning diffeomorphic deformations, it efficiently generates accurate white matter and pial surfaces, improving geometric fidelity and morphometric reliability across datasets for robust surface ...
Chengjin Li   +10 more
wiley   +1 more source

Enhancing Super‐Resolution Spatial Transcriptomics Data by Transfer Learning

open access: yesAdvanced Science, EarlyView.
SpotZoomer employs a transfer‐learning‐based strategy to enhance the resolution of Visium data by leveraging available high‐resolution priors. The resulting super‐resolved maps enable sharper delineation of cell boundaries and more precise inference of cell–cell communication patterns that would otherwise remain obscured at native resolution.
Xiaoyu Li, Lihua Zhang, Wenwen Min
wiley   +1 more source

Human‐Guided Bayesian Optimization Enables High‐Throughput Laser Annealing of Mesoporous SiOx Anodes for Lithium‐Ion Batteries

open access: yesAdvanced Science, EarlyView.
An empirical‐aided active learning framework is developed to optimize high‐throughput laser‐induced photothermal annealing of silicon suboxide anodes. By integrating probabilistic machine learning with empirical domain knowledge, this approach achieves optimal electrochemical performance using limited experiments.
Chaeyoung Park   +3 more
wiley   +1 more source

Gap‐Free Information Transfer in 4D‐STEM via Fusion of Complementary Scattering Channels

open access: yesAdvanced Science, EarlyView.
Fused Full‐Field STEM (FF‐STEM) is introduced as a 4D‐STEM imaging modality that combines direct ptychography with tilt‐corrected dark‐field reconstruction in a single acquisition. Fourier‐space fusion using Wiener‐type spectral weighting closes the low‐frequency contrast gap inherent to bright‐field methods, delivering gap‐free, dose‐efficient, near ...
Shengbo You   +15 more
wiley   +1 more source

Machine Learning–Guided Surface Strain Engineering in Connected Platinum–Nickel Nanoparticle Catalysts for Advanced Oxygen Reduction Performance

open access: yesAdvanced Science, EarlyView.
Machine learning‐guided strain engineering enables highly active, durable, support‐free Pt–Ni nanonetwork catalysts for the oxygen reduction reaction. Analysis of a Pt‐based catalyst dataset identifies surface compressive strain as an effective descriptor associated with enhanced activity and provided practical design guidelines.
Aparna Chitra Sudheer   +4 more
wiley   +1 more source

UCtracker: A Deep Learning–Based DNA Methylation Model for Noninvasive Diagnosis and Recurrence Surveillance of Urothelial Carcinoma in a Prospective Study

open access: yesAdvanced Science, EarlyView.
We developed UCtracker, a urine DNA methylation–based deep learning model, for noninvasive diagnosis and postoperative surveillance of urothelial carcinoma. UCtracker demonstrates high diagnostic accuracy, robustness at ultralow sequencing depth, early recurrence detection, and dynamic risk‐stratified monitoring of molecular residual disease ...
Shengwei Xiong   +19 more
wiley   +1 more source

Taming hyperparameter tuning in continuous normalizing flows using the JKO scheme. [PDF]

open access: yesSci Rep, 2023
Vidal A   +4 more
europepmc   +1 more source

Unveiling Mechanisms of SEI Formation and Sodium Loss in Sodium Batteries via Interface Reactor Sampling

open access: yesAdvanced Science, EarlyView.
An “Interface Reactor” strategy boosts simulation stability by 2–3 orders of magnitude, enabling stable 100 ns molecular dynamics of electrode‐electrolyte interfaces. Distinct SEI formation mechanisms are revealed: mixed co‐formation in carbonates versus surface‐energy‐controlled NaF crystallization in ethers. Metadynamics simulations further elucidate
Zhoulin Liu   +6 more
wiley   +1 more source

CLRe: A Synergistic Dual‐Engine Framework for One‐Step Retrosynthesis Prediction

open access: yesAdvanced Science, EarlyView.
CLRe uses a contrastive difficulty score to order pretrained seq2seq fine‐tuning for retrosynthesis. Reaction embeddings define the ranking score, and a cumulative easy‐to‐hard schedule expands from the easiest subset to the full training set while earlier examples remain active.
Tianhao Su   +5 more
wiley   +1 more source

Home - About - Disclaimer - Privacy