Unsupervised domain adaptation enables neural networks to transfer from a labeled source domain to an unlabeled target domain by learning domain-invariant representations.Recent approaches achieve this by directly mat...Unsupervised domain adaptation enables neural networks to transfer from a labeled source domain to an unlabeled target domain by learning domain-invariant representations.Recent approaches achieve this by directly matching the marginal distributions of these two domains.Most of them,however,ignore exploration of the dynamic trade-off between domain alignment and semantic discrimination learning,thus rendering them susceptible to the problems of negative transfer and outlier samples.To address these issues,we introduce the dynamic parameterized learning framework.First,by exploring domain-level semantic knowledge,the dynamic alignment parameter is proposed,to adaptively adjust the optimization steps of domain alignment and semantic discrimination learning.Besides,for obtaining semantic-discriminative and domain-invariant representations,we propose to align training trajectories on both source and target domains.Comprehensive experiments are conducted to validate the effectiveness of the proposed methods,and extensive comparisons are conducted on seven datasets of three visual tasks to demonstrate their practicability.展开更多
As an emerging two-dimensional(2D)semiconductor material,monolayer MoS2 has recently attracted considerable attention.Various promising applications of this material have been proposed for electronics,optoelectronics,...As an emerging two-dimensional(2D)semiconductor material,monolayer MoS2 has recently attracted considerable attention.Various promising applications of this material have been proposed for electronics,optoelectronics,sensing,catalysis,energy storage,and so on.To realize these practical applications,high-quality and large-area MoS2 with controllable properties is required.Among the many different synthesis techniques,epitaxy provides a promising route for producing MoS2 monolayers.Here,we review the epitaxial growth of monolayer MoS2 on various substrates,with a particular focus on large-scale films with large domain sizes and high domain alignments.Finally,we offer perspectives and challenges for future research and applications of this technology.展开更多
基金supported by the National Natural Science Foundation of China(No.61932009)the Starry Night Science Fund of Zhejiang University Shanghai Institute for Advanced Study,China。
文摘Unsupervised domain adaptation enables neural networks to transfer from a labeled source domain to an unlabeled target domain by learning domain-invariant representations.Recent approaches achieve this by directly matching the marginal distributions of these two domains.Most of them,however,ignore exploration of the dynamic trade-off between domain alignment and semantic discrimination learning,thus rendering them susceptible to the problems of negative transfer and outlier samples.To address these issues,we introduce the dynamic parameterized learning framework.First,by exploring domain-level semantic knowledge,the dynamic alignment parameter is proposed,to adaptively adjust the optimization steps of domain alignment and semantic discrimination learning.Besides,for obtaining semantic-discriminative and domain-invariant representations,we propose to align training trajectories on both source and target domains.Comprehensive experiments are conducted to validate the effectiveness of the proposed methods,and extensive comparisons are conducted on seven datasets of three visual tasks to demonstrate their practicability.
基金the National Natural Science Foundation of China(Nos.11834017 and 61888102)the National Key Research and Development Program of China(No.2016YFA0300904)+1 种基金the Key Research Program of Frontier Sciences of CAS(No.QYZDB-SSW-SLH004)the Strategic Priority Research Program of CAS(Nos.XDB30302000 and XDB33010300).
文摘As an emerging two-dimensional(2D)semiconductor material,monolayer MoS2 has recently attracted considerable attention.Various promising applications of this material have been proposed for electronics,optoelectronics,sensing,catalysis,energy storage,and so on.To realize these practical applications,high-quality and large-area MoS2 with controllable properties is required.Among the many different synthesis techniques,epitaxy provides a promising route for producing MoS2 monolayers.Here,we review the epitaxial growth of monolayer MoS2 on various substrates,with a particular focus on large-scale films with large domain sizes and high domain alignments.Finally,we offer perspectives and challenges for future research and applications of this technology.
