Yonghui Cong

37 XueYuan Rd, Beihang University (BUAA), Beijing, P.R.China, 100191 

mailto:zslibuaa@gmail.com•+86-134-2600-3202

 

EDUCATION

Sep. 2012 ~ Present  

M.S. in Pattern Recognition, School of Automation Science and Electrical Engineering ,Beihang University,Beijing, P.R.China

 

Sep.2008 ~ Jul.2012 

 B.E. in Automation, Minzu University
School of Information and Engineering, Beijing, P.R.China


RESEARCH INTEREST

Natural Language Processing , Cross-modal Retrieval, Machine Learning


RESEARCH

Cross-modal retrieval system based on Chinese Database

  • Contribution: Apply the probabilistic model using the topic correlation model for cross-modal information retrieval in different modalities such as images and text. SVM classifiers are applied to compute the probability distribution over categories. Mining potential semantic relations between the different modalities and achieving higher efficiency for cross-modal retrieval. Primarily responsible for the experiments on the English database. Established a Chinese database with images and texts and carried out cross-modal testing.
  • Framework: Extract the sift features of the images and use of the LDA topic model of texts for train→ Use SVM classifiers to compute the probability distribution over categories→ Get the sift features of a test image or the topic distribution of a test text→ Get a retrieval result ranked by semantic closeness.

Text POS tagging based on semantics and syntax information

  • Framework: We are interested in a generative model of text associated with two latent variables: semantic topics and syntax classes. Combining syntax and semantics into a coherent probabilistic generative model for texts tagging. It is an extension and generation of LDA and Bayes-HMM that is designed to understand the short-range dependencies between words. The LDA model is used to get the topic information while the syntax classes are generated through a Bayesian HMM. Posterior inference is used to solve the problem in topic and syntax models.

PUBLICATIONS

  • Jing Yu, Yonghui Cong, Zengchang Qin, and Tao Wan (2012), Cross-Modal Topic Correlations for Multimedia Retrieval,  ICPR 2012.
  • Yonghui Cong, Zengchang Qin, Jing Yu and Tao Wan (2012), Cross-Modal Information Retrieval - A Case Study on Chinese Wikipedia, ADMA 2012.
  • Zengchang Qin, Jing Yu, Yonghui Cong and Tao Wan(2013), Topic Correlation Model for Cross-modal Multimedia Information Retrieval, Submit to PAA.

HONORS AND AWARDS

  • 2012-Excellent graduate in Beijing
  • 2012-The Baosteel outstanding student award
  • 2011-Miyoshi students of Beijing
  • 2008~2012-Scholarship for Excellent Academic Performance (4 times) Minzu University
  • 2011-The second prize of Mathematical Modeling  Minzu University
  • 2011-Third Prize in National “TI” Cup Internet of Things Contest
  • 2009-The National Scholarship Minzu University

INTERNSHIP EXPERIENCE

  • Jul.2012 – Sep.2012   Siemens (China) CT

SKILLS

  • C,C++,matlab,python


Copyleft © 2011

Valid HTML 4.01 Transitional