Results 91 to 100 of about 259,517 (255)

Alpha MAML: Adaptive Model-Agnostic Meta-Learning

open access: yes, 2019
Model-agnostic meta-learning (MAML) is a meta-learning technique to train a model on a multitude of learning tasks in a way that primes the model for few-shot learning of new tasks.
Baydin, Atılım Güneş   +2 more
core  

Weighted Meta-Learning

open access: yes, 2020
18 pages, 7 ...
Cai, Diana   +3 more
openaire   +2 more sources

Recomposable Layered Metasurfaces for Wavelength‐Multiplexed Optical Encryption via Modular Diffractive Deep Neural Networks

open access: yesAdvanced Functional Materials, EarlyView.
Modular diffractive deep neural network metasurfaces encode and reconstruct holograms across layer combinations and wavelengths, enabling secure, multifunctional operation. Each layer acts independently yet composes jointly, yielding up to m(2N −1) channels for m wavelengths and N layers.
Cherry Park   +4 more
wiley   +1 more source

Artificial Intelligence as the Next Visionary in Liquid Crystal Research

open access: yesAdvanced Functional Materials, EarlyView.
The functions of AI in the research laboratory are becoming increasingly sophisticated, allowing the entire process of hypothesis formulation, material design, synthesis, experimental design, and reiterative testing to be automated. In our work, we conceive how the incorporation of AI in the laboratory environment will transform the role and ...
Mert O. Astam   +2 more
wiley   +1 more source

Deep Learning-Based Maximum Temperature Forecasting Assisted with Meta-Learning for Hyperparameter Optimization

open access: yesAtmosphere, 2020
Time series forecasting of meteorological variables such as daily temperature has recently drawn considerable attention from researchers to address the limitations of traditional forecasting models.
Trang Thi Kieu Tran   +4 more
doaj   +1 more source

Hierarchical Meta Learning

open access: yes, 2019
Meta learning is a promising solution to few-shot learning problems. However, existing meta learning methods are restricted to the scenarios where training and application tasks share the same out-put structure. To obtain a meta model applicable to the tasks with new structures, it is required to collect new training data and repeat the time-consuming ...
Zou, Yingtian, Feng, Jiashi
openaire   +2 more sources

The Cuttlebone Blueprint for Multifunctional Metamaterials: Design Taxonomy, Functional Decoupling, and Future Horizons

open access: yesAdvanced Functional Materials, EarlyView.
Cuttlebone‐inspired metamaterials exploit a septum‐wall architecture to achieve excellent mechanical and functional properties. This review classifies existing designs into direct biomimetic, honeycomb‐type, and strut‐type architectures, summarizes governing design principles, and presents a decoupled design framework for interpreting multiphysical ...
Xinwei Li, Zhendong Li
wiley   +1 more source

Evaluation of the Dual Impact of Nanotechnologies on Health and Environment Through Alternative Bridging Models

open access: yesAdvanced Healthcare Materials, EarlyView.
This review explores how alternative invertebrate and small‐vertebrate models advance the evaluation of nanomaterials across medicine and environmental science. By bridging cellular and organismal levels, these models enable integrated assessment of toxicity, biodistribution, and therapeutic performance.
Marie Celine Lefevre   +3 more
wiley   +1 more source

Beyond Presumptions: Toward Mechanistic Clarity in Metal‐Free Carbon Catalysts for Electrochemical H2O2 Production via Data Science

open access: yesAdvanced Materials, EarlyView.
Metal‐free carbon catalysts enable the sustainable synthesis of hydrogen peroxide via two‐electron oxygen reduction; however, active site complexity continues to hinder reliable interpretation. This review critiques correlation‐based approaches and highlights the importance of orthogonal experimental designs, standardized catalyst passports ...
Dayu Zhu   +3 more
wiley   +1 more source

Online Meta-Learning

open access: yes, 2019
ICML 2019. The first two authors contributed equally.
Finn, Chelsea   +3 more
openaire   +2 more sources

Home - About - Disclaimer - Privacy