Results 141 to 150 of about 2,019,981 (302)

Beyond Manual Tuning of Hyperparameters

, 2015
The success of hand-crafted machine learning systems in many applications raises the question of making machine learning algorithms more autonomous, i.e., to reduce the requirement of expert input to a minimum. We discuss two strategies towards this goal:
Hutter, Frank, Schmidt-Thieme, Lars, Lücke, Jörg   +2 more
core   +1 more source

High‐Throughput Data Generation and Transfer Learning Enabled Microstructure‐Property Integrated Design of Nickel‐Based Powder Metallurgy Superalloy

Advanced Science, EarlyView.
An integrated transfer learning framework integrates CALPHAD simulations, diffusion‐multiple experiments, and literature data to predict long‐term microstructural stability and short‐term mechanical properties of Ni‐based powder metallurgy superalloys. Based on these model predictions, a high‐performance, low‐density alloy, USTB‐PM750, is designed from
Zixin Li, Hongtao Zhang, Zichao Peng, Yuheng Zhang, Qian Lv, Yihuan Cao, Xuqing Wang, Huadong Fu, Jianxin Xie   +8 more
wiley   +1 more source

A Hybrid Optimization Model for Transformer Fault Diagnosis Based on Gas Classification

Digital
Dissolved gas analysis (DGA) provides valuable information for transformer condition monitoring, yet accurate multi-class fault identification remains challenging due to overlapping gas patterns and the sensitivity of classifier hyperparameters.
Junju Lai, Dongpeng Weng, Feng Xian, Yuandong Xie, Yujie Chen, Qian Zhou, Chao Yuan   +6 more
doaj   +1 more source

Dynamic Priors in Bayesian Optimization for Hyperparameter Optimization

CoRR
Bayesian optimization (BO) is a widely used approach to hyperparameter optimization (HPO). However, most existing HPO methods only incorporate expert knowledge during initialization, limiting practitioners' ability to influence the optimization process as new insights emerge. This limits the applicability of BO in iterative machine learning development
Lukas Fehring, Marcel Wever, Maximilian Spliethöver, Leona Hennig, Henning Wachsmuth, Marius Lindauer   +5 more
openaire   +2 more sources

Understanding weight-magnitude hyperparameters in training binary networks

, 2022
Binary Neural Networks (BNNs) are compact and efficient by using binary weights instead of real-valued weights. Current BNNs use latent real-valued weights during training, where several training hyper-parameters are inherited from real-valued networks ...
Quist, Joris (author)
core  

Deciphering Short‐Range Order in 2D Transition Metal Dichalcogenides: From Origin to Multi‐Scale Property Modulation

Advanced Science, EarlyView.
Short‐range order in 2D transition metal dichalcogenides is revealed as a new design paradigm. Driven by chemical affinity and atomic size, it governs properties across scales. Weak ordering tunes site‐resolved magnetism and d‐band centers, while strong ordering eliminates gap states to open band gaps.
Hanyu Liu, Linggang Zhu, Jian Zhou, Zhimei Sun   +3 more
wiley   +1 more source

Efficient Screening of Organic Singlet Fission Molecules Using Graph Neural Networks

Advanced Science, EarlyView.
A high‐throughput screening framework based on graph neural networks (GNNs) and multi‐level validation facilitates the identification of singlet fission (SF) candidates. By efficiently predicting excitation energies across 20 million molecules, and integrating TDDFT calculations, synthetic accessibility assessments, and GW+BSE calculations, this ...
Li Fu, Longfei Lv, Fan Zhang, Si Zhou, Weiwei Gao, Jijun Zhao   +5 more
wiley   +1 more source

Learning General Gaussian Kernel Hyperparameters for SVR

, 2013
International audienceWe propose a new method for general gaussian kernel hyperparameters optimization for support vector regression. The hyperparameters are constrained to lie on a differentiable manifold. The proposed optimization technique is based on
Snoussi, Hichem, Abdallah, Fahed, R. Lengellé, F. Abdallah, Laanaya, Hicham, Lengellé, Régis, H. Laanaya, Hichem Snoussi   +7 more
core   +1 more source

High‐Throughput Screening and Interpretable Machine Learning for Rational Design of Bimetallic Catalysts for Methane Activation

Advanced Science, EarlyView.
ABSTRACT Methane's efficient catalytic removal is vital for sustainable development. Bimetallic catalysts, though promising for methane activation, pose a design challenge due to their complex compositional space. This work introduces an integrated framework that combines high‐throughput density functional theory (DFT) and interpretable machine ...
Mingzhang Pan, Tian Zhang, Jiawei Dong, Yubao Xie, Kaifeng Zhong, Wei Guan, Haiqiao Wei, Changcheng Fu, Yaqiong Su   +8 more
wiley   +1 more source

Data‐Driven Modeling of Composition–Processing–Microstructure Relations for Recycled Aluminum Cast Alloys

Advanced Science, EarlyView.
Interpretable machine learning reveals how composition and processing govern the formation and microstructural burden of Fe‐rich intermetallic compounds in recycled Al–Si–Fe–Mn alloys. By separating morphology selection from morphology‐conditioned burden partitioning, this framework shows that identical Fe contents can yield different intermetallic ...
Jaemin Wang, Waleed Mohammed, Dierk Raabe   +2 more
wiley   +1 more source