A5053669769- Peer-to-Peer Network Technologies
- Caching and Content Delivery
- Optical Coherence Tomography Applications
- Advanced Fluorescence Microscopy Techniques
- Topic Modeling
- Semantic Web and Ontologies
- Advanced Data Storage Technologies
- Photoacoustic and Ultrasonic Imaging
- Data Quality and Management
- Advanced Fiber Optic Sensors
- Recommender Systems and Techniques
- Handwritten Text Recognition Techniques
- Multimodal Machine Learning Applications
- Image Retrieval and Classification Techniques
- Mobile Ad Hoc Networks
- Natural Language Processing Techniques
- Semiconductor Lasers and Optical Devices
- Opportunistic and Delay-Tolerant Networks
- Advanced Database Systems and Queries
East China Normal University
2020
Beijing University of Technology
2012-2014
Fudan University
2004-2005
Recently most handwritten mathematical expression recognition methods adopt the attention-based encoder-decoder framework, which generates LaTeX sequences from given images. However, accuracy of attention mechanism limits performance HMER models. Lacking global context information in decoding process is also a challenge for HMER. Some symbol-level counting to localize symbols improving model performance, while these cannot work well. In this paper, we propose method named SLAN, shorted...
In this paper, we present C2, a new overlay network based on CAN and Chord. It is primarily designed for dynamic environment in which peers join depart the frequently. For an n-peers C2 system, each peer maintains only about O(log n) of other peers' information, achieves routing within hops. peer's joining or departure, C can, high probability, update tables with no more than messages. What distinguishes from many peer-to-peer data sharing systems its low computation cost efficiency network....
Author disambiguation arises when different authors share the same name, which is a critical task in digital libraries, such as DBLP, CiteULike, CiteSeerX, etc. While state-of-the-art methods have developed various paper embedding-based performing top-down manner, they primarily focus on ego-network of target name and overlook low-quality collaborative relations existed ego-network. Thus, these can be suboptimal for disambiguating authors.In this paper, we model author collaboration network...
We propose and experimentally demonstrate a shift-multiplexing complex spectral-domain optical coherence tomography (shift-multiplexing CSD-OCT) method, in which the maximum detection depth of SD-OCT can be greatly extended by incorporating positions with CSD-OCT. The tomographic imaging twofold or threefold microscopic slides as target sample is performed. experimental results show that more uniform brightness clarity for different regions thick achieved CSD-OCT system. In particular, even...
Optical Coherence Tomography (OCT) was successfully applied in the microstructure imaging of biological tissue after being proposed firstly 1991 by researchers MIT. As a novel optical technology, it mainly uses interference principles to achieve noninvasive and high resolution visualization samples. OCT works analogously an ultrasound scanner, major difference is that pulses are replaced broadband light. According whether need for mechanical axial scan depth direction, can be classified into...
The three-step or many steps phase shifting method is usually employed to resolve the complex-conjugate ambiguity in Spectral-domain optical coherence tomography (SD-OCT). However it reduces image quality and also imaging speed slow. In this paper two phase-shifting used digital processing improves of reconstructed SD-OCT. two-step operation only once which simplified experiment effect relative error SD-OCT on eliminated.
Author disambiguation arises when different authors share the same name, which is a critical task in digital libraries, such as DBLP, CiteULike, CiteSeerX, etc. While state-of-the-art methods have developed various paper embedding-based performing top-down manner, they primarily focus on ego-network of target name and overlook low-quality collaborative relations existed ego-network. Thus, these can be suboptimal for disambiguating authors. In this paper, we model author collaboration network...
Structural image content recognition (SICR) aims to transcribe a two-dimensional structural (e.g., mathematical expression, chemical formula, or music score) into token sequence. Existing methods are mainly encoder-decoder based and overlook the importance of feature selection spatial relation extraction in map. In this paper, we propose DEAL (shorted for Dynamic fEAture seLection) SICR, which contains dynamic selector extractor as two cornerstone modules. Specifically, novel loss function...