检索结果-南通市图书馆

Contrastive Self-supervised Representation Learning Using Synthetic Data

在线全文

学校读者我要写书评

暂无评论

International Journal of Automation and computing 2021年第4期18卷 556-567页

作者： Dong-Yu she.kun Xu Beijing National Research Center for Information Science and Technology Department of Computer Science and TechnologyTsinghua UniversityBeijing 100084China

Learning discriminative representations with deep neural networks often relies on massive labeled data, which is expensive and difficult to obtain in many real scenarios. As an alternative, self-supervised learning that leverages input itself as supervision is strongly preferred for its soaring performance on visual representation learning. This paper introduces a contrastive self-supervised framework for learning generalizable representations on the synthetic data that can be obtained easily with complete ***, we propose to optimize a contrastive learning task and a physical property prediction task simultaneously. Given the synthetic scene, the first task aims to maximize agreement between a pair of synthetic images generated by our proposed view sampling module, while the second task aims to predict three physical property maps, i.e., depth, instance contour maps, and surface normal maps. In addition, a feature-level domain adaptation technique with adversarial training is applied to reduce the domain difference between the realistic and the synthetic data. Experiments demonstrate that our proposed method achieves state-of-the-art performance on several visual recognition datasets.

关键词： Self-supervised learning contrastive learning synthetic image convolutional neural network representation learning

Fabrication and Life Prediction of SSiC Ceramic Joint Joined with Silicon Resin YR3370

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Aeronautics 2007年第1期20卷 81-85页

作者： YUAN Xiao-kun XU Bing-she School of Materials Science and Engineering Beijing University of Aeronautics and Astronautics Beijing 100083 China College of Materials Science and Engineering Taiyuan University of Technology Taiyuan 030024 China

Joints between sintered silicon carbide （SSiC） were produced using a polysiloxane silicon resin YR3370 （GE Toshiba Silicones） as joining material. Samples were heat treated in a 99.99% nitrogen flux at temperatures ranging from 1 100 ℃ to 1 300 ℃. Three point bending strength of the joint reached the maximum of 179 MPa as joined at 1 200℃. The joining layer is continuous, homogeneous and densified and has a thickness of 2 μm -5μm. The joining mechanism is that the amorphous silicon oxycarbide （SixOyCz） ceramic pyrolyzed from silicon resin YR3370 acts as an inorganic adhesive to SSiC substrate, which means the formation of the continuous Si-C bond structure between SixOyCz structure and SSiC substrate. Life prediction of the ceramic joint can be realized through the measurement of the critical time of the joint after the cyclic loading test.

关键词： silicon resin YR3370 SiC ceramic ceramic joining joint life

Overlap digraph:An effective model for finding good spaced seeds for biological sequence local alignment

在线全文

学校读者我要写书评

暂无评论

Chinese Science Bulletin 2011年第11期56卷 1100-1107页

作者： CHEN Ke she.kun ZHU QingXin School of Computer University of Electronic Science and Technology of China Chengdu 610054 China

Spaced seeds technology,which was proposed by PatternHunter,has been proven to be more sensitive and faster than continuous seeds,and it is now widely used for bio-sequence local ***,finding optimal spaced seeds is an NP-hard problem.A seed digraph model is proposed to find good spaced seeds,which are very close to optimal,in a very different but effective *** this different approach,some good long spaced seeds which cannot be calculated by normal optimal sensitivity formulas due to their exponential complexity can be found.

关键词：种子技术有向图模型生物序列种用灵敏度长间隔复杂性

Interfacial Characteristic of as-Deformed SiCp-Reinforced Magnesium Matrix Composite

在线全文

学校读者我要写书评

暂无评论

Acta Metallurgica Sinica(English Letters) 2014年第5期27卷 885-893页

作者： kun kun Deng Jian Chao Li Jian Feng Fan Xiao Jun Wang kun Wu Bing she.Xu College of Materials Science and Engineering Taiyuan University of Technology Key laboratory of interface science and engineering in advanced materials Ministry of EducationTaiyuan University of Technology School of Materials Science and Engineering Harbin Institute of Technology

The （submicron ＋ micron） SiCp-reinforced magnesium matrix composite was fabricated by stir casting. After the application of forging and extrusion, the interface between SiCp and Mg in the composite was investigated by transmission electron microscopy. Results show that the interfacial characterization was different at the interfaces of micron-SiCp/Mg and submicron-SiCp/Mg. While most interfaces between micron-SiCp and Mg were clean, the precip- itated Mg17A112 phase as well as dispersedly distributed nano-MgO particles was observed at some interfaces. Unlike the interface between micron-SiCp and Mg, no interfacial reaction product was found at the interface between submicron-SiCp and Mg in the present study. Besides, the specific orientation relationships were found at the interfaces between submicron- SiCp and Mg, which was thought to have developed during hot deformation process. At the fracture surface of the composite, the microcracks were found at the interface between micron-SiCp and Mg, while the interfacial bonding between submicron-SiCp and Mg was very well.

关键词： Magnesium matrix composite SiC particles Interfaces TEM

Lithium intercalation mechanism for β-SnSb in Sn-Sb thin films

在线全文

学校读者我要写书评

暂无评论

International Journal of Minerals,Metallurgy and Materials 2011年第2期18卷 216-222页

作者： Qiang Ru Qin Tian she.jun Hu ling-zhi Zhao Laboratory of Quantum Information Technology School of Physics and Telecommunication Engineering South China Normal University Guangzhou 510006 China Department of Chemistry South China Normal University Guangzhou 510006 China

Based on the first-principles plane wave pseudo-potential method, the electronic structure and electrochemical performance of LixSn4Sb4 （x=2, 4, 6, and 8） and LixSn12-xSb4 （x=9, 10, 11, and 12） phases were calculated. A Sn-Sb thin film on a Cu foil was also prepared by radio frequency magnetron sputtering. The surface morphology, composition, and lithium intercalation/extraction behavior of the fabricated film were characterized by scanning electron microscopy （SEM）, X-ray diffraction （XRD） and cyclic voltammetry （CV）. Lithium atoms can easily insert into and extract out of the β-SnSb cell due to the low lithium intercalation formation energy. It is found that lithium atoms first occupy the interstitial sites, and then Sn atoms at the lattice positions are replaced by excessive lithium. The dissociative Sn atoms continue to produce different Li-Sn phases, which will affect the electrode stability and lead to the undesirable effect due to their large volume expansion ratio. The calculated lithium intercalation potential is stable at about 0.7 V, which is consistent with the experimental result.

关键词： lithium batteries first-principles Sn-Sb alloys electronic structure thin films

在线全文

学校读者我要写书评

暂无评论

距骨骨软骨损伤治疗方案的个性化组合

中国组织工程研究 2025年第21期29卷 4584-4592页

作者：陈学明佘昶无锡市第九人民医院(苏州大学附属无锡九院)足踝外科江苏省无锡市214000 苏州大学附属第二医院骨科江苏省苏州市215004

背景:距骨骨软骨损伤根据不同分型或临床症状,采取不同的治疗方法,包括保守治疗、手术治疗和生物制剂治疗。保守治疗大多可短时缓解症状和延缓病情;近年来关节镜下微骨折、克氏针钻孔、距骨软骨移植、自体骨软骨细胞移植、干细胞移植和... 详细信息

背景:距骨骨软骨损伤根据不同分型或临床症状,采取不同的治疗方法,包括保守治疗、手术治疗和生物制剂治疗。保守治疗大多可短时缓解症状和延缓病情;近年来关节镜下微骨折、克氏针钻孔、距骨软骨移植、自体骨软骨细胞移植、干细胞移植和生物制剂富血小板血浆等方法,都取得了不错的疗效。目的:对距骨骨软骨损伤的治疗进展进行总结,为其临床治疗提供参考。方法:第一作者以“talus,osteochondral lesions of the talus,微骨折,移植术,富血小板血浆,干细胞”为检索词,检索Pub Med数据库及中国知网2019年1月至2024年1月的相关文章。共纳入67篇文献进行综合分析。结果与结论:(1)目前治疗距骨骨软骨损伤的方式较多,对于Hepple分型Ⅰ和Ⅱ型病变保守康复治疗和生物制剂可缓解病情;Hepple分型Ⅲ型及以上且病变直径<15 mm选择微骨折或克氏针钻孔骨髓刺激治疗,但长期疗效不佳;更大病变面积选择置换和再生策略,移植失败是其最主要风险;(2)生物制剂治疗作为一种治疗新方式,多用于组合治疗;(3)针对距骨骨软骨损伤的治疗,需要考虑到软骨损伤的面积、部位等诸多因素,其治疗方案的选择还应考虑医疗技术开展情况以及患者接受程度、经济状况等因素,建议对各种方案进行个性化组合治疗,以取得更为良好的效果。

关键词：距骨距骨骨软骨损伤微骨折移植术富血小板血浆干细胞

A Novel Visualization Tool for Manual Annotation when Building Large Speech Corpora

在线全文

学校读者我要写书评

暂无评论

Wuhan University Journal of Natural Sciences 2006年第2期11卷 381-384页

作者： she.kun CHEN Shuzhen YANG she. ZOU Lian School of Electronic Information Wuhan University Wuhan 430072 Hubei China

A novel visualized sound description, called sound dendrogram is proposed to make manual annotation easier when building large speech corpora. It is a lattice structure built from a group of ＂seed regions＂ and through an iterative procedure of mergence. A simple but reliable extraction method of ＂seed regkms＂ and advanced distance metric are adopted to construct the sound dendrogram, so that it can present speech＇s structure character ranging from coarse to fine in a visualized way. Tests show that all phonemic boundaries are contained in the lattice structure of sound dendrogram and very easy to identify. Sound dendrogram can be a powerful assistant tool during the process of speech corporals manual annotation.

关键词： sound dedrogram speech corpora manual annotation computer aid tool

Micromagnetic studies of perpendicular recording FePt media with controllable grain size distributions

在线全文

学校读者我要写书评

暂无评论

Rare Metals 2006年第Z1期25卷 436-439页

作者： DONG Zhong PIAO kun she.she.gxian WEI Dan LI Zhenghua WEI Fulin Department of Materials Science and Engineering Tsinghua UniversityBeijing 100084China Research Institute of Magnetic Materials Lanzhou UniversityLanzhou 730000China

A 3-dimensional(3D)micromagnetic model combined with Fast Fourier Transform(FFT)method was built up to study the writability in the L1_(0)FePt perpendicular *** effects of controllable grain size distributions were studied by grain growth *** is found that the cross-track-averaged magnetization changes little between the L1_(0)FePt medium with uniform or non-uniform grain size distribution.

关键词： perpendicular recording L1_(0)FePt medium controllable grain size distributions micromagnetic simulation

Tuning-up Learning Parameters for Deep Convolutional Neural Network:A Case Study for Hand-Drawn Sketch Images

在线全文

学校读者我要写书评

暂无评论

Journal of Electronic Science and Technology 2022年第3期20卷 305-318页

作者： Shaukat Hayat kun she.Muhammad Mateen Parinya Suwansrikham Muhammad Abdullah Ahmed Alghaili the School of Information and Software Engineering University of Electronic Science and Technology of ChinaChengdu 610054 the School of Big Data&Software Engineering Chongqing UniversityChongqing 400044 the College of Information Science and Engineering Hunan UniversityChangsha 410082

Several recent successes in deep learning(DL),such as state-of-the-art performance on several image classification benchmarks,have been achieved through the improved ***(HPs)tuning is a key factor affecting the performance of machine learning(ML)*** state-of-the-art DL models use different HPs in different ways for classification tasks on different *** manuscript provides a brief overview of learning parameters and configuration techniques to show the benefits of using a large-scale handdrawn sketch dataset for classification *** analyzed the impact of different learning parameters and toplayer configurations with batch normalization(BN)and dropouts on the performance of the pre-trained visual geometry group 19(VGG-19).The analyzed learning parameters include different learning rates and momentum values of two different optimizers,such as stochastic gradient descent(SGD)and *** analysis demonstrates that using the SGD optimizer and learning parameters,such as small learning rates with high values of momentum,along with both BN and dropouts in top layers,has a good impact on the sketch image classification accuracy.

关键词： Deep learning(DL) hand-drawn sketches learning parameters