I am a PhD student in the College of Computer and Information Science
(CCIS) of Northeastern
University. I graduated from Zhejiang Sci-Tech University in Hangzhou, China
with a Bachelor's degree in computer science and technology. I was
born in Ningbo, and grew up in Hangzhou, a beautiful ancient city in east
China.
Now I am working with Prof. Javed Aslam on information retrieval
and machine learning. Meanwhile, I am also working with Prof. Harriet Fell on developing speech
processing toolkit for speech pathologists. When I am not
doing research, I like to spend my time entertaining my cats
Minnie & Zora, photographing, hiking, and
watching movies. Welcome to visit my photo blog!
Research
I am currently working on modeling score distribution for relevant and non-relevant documents and query categorization for learning-to-rank. In modeling score distribution, I am studying what are the appropriate statistical distributions to model relevant and non-relevant document scores returned by certain information retrieval system, and how to utilize the score distribution to predict the retrieval system performance and improve the ranking.
In learning to rank, I am investigating query-dependent ranking based on training different ranking functions over different categories of queries. I am also responsible for extracting document features and building training/testing dataset for learning algorithms. In summer 2009, I participated in building a learning-to-rank framework based information retrieval system for Million Query track of Text REtrieval Conference (TREC) 2009.
Before these, I have also done following projects on speech processing for assistive technology in human computer interaction:
- Emotion in Speech: recognizing different emotions in speech vocalizations, and comparing emotions using acoustic features and human perceptional dimensions.
- visiBabble: a system that processes vocalizations and responds with visual feedbacks in real-time to encourage the pre-speech vocalization in children who are at risk of impaired speech development.
Publications
Keshi Dai, Virgil Pavlu, Evangelos Kanoulas, and Javed Aslam, Extended Expectation Maximization for Inferring Scoreb Distributions, in Proc. of ECIR 2012: Advances in Information Retrieval: 34th European Conference on IR Research. [bibTex] [pdf]
Evangelos Kanoulas, Keshi Dai, Virgil Pavlu, and Javed Aslam, Score Distribution Models: Assumptions, Intuition, and Robustness to Score Manipulation, in Proc. of the 33rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 2010, Geneva, Switzerland. [bibTex] [pdf]
Keshi Dai, Evangelos Kanoulas, Virgil Pavlu, and Javed Aslam, Variational Bayes for Modeling Score Distribution, in Journal of Information Retrieval, ICTIR Special Issue, 2010 [bibTex] [pdf]
Evangelos Kanoulas, Virgil Pavlu, Keshi Dai, and Javed Aslam, Modeling the Score Distributions of Relevant and Non-relevant Documents, in Proc. of the 2nd International Conference on the Theory of Information Retrieval (ICTIR), 2009, Cambridge, UK. [bibTex] [pdf]
Keshi Dai, Harriet J. Fell, and Joel MacAuslan, Comparing Emotions using Acoustics and Human Perceptional Dimensions in Proc. of the 27th International Conference Extended Abstracts on Human Factors in Computing Systems (CHI), 2009, Boston, USA. [bibTex] [pdf]
Keshi Dai, Harriet J. Fell, and Joel MacAuslan, Recognizing Emotion in Speech Using Neural Networks, in Proc. of the IASTED International Conference on Telehealth and Assistive Technologies, Baltimore, 2008, MD, USA. [bibTex] [pdf]
Harriet J. Fell, Joel MacAuslan, Jun Gong, Keshi Dai, et al., VisiBabble: a System for Reinforcement of Early Vocalization, poster for the American Speech-Language-Hearing Association's Annual Convention, 2007, Boston, MA, USA.
Keshi Dai, Early Diagnosis of Autism through Analysis of Pre-speech Vocalization, Journal of ACM SIGACCESS Accessibility and Computing, 89:42-46, 2007. [bibTex] [pdf]
Notes & Codes
Variational Bayesian Inference: [note]
Expectation Maximization: Gaussian-Exponential mixture [matlab]; Gamma-Gaussian mixture [matlab]; k-Gaussian mixture [matlab]
Audio Show: [matlab]