|
|
Past Projects  | Three-Layer Design for Distributed Voting System
Fall 2006 CS271 Course Project, with Chang-Ming Tsai
We implemented a distributed electronic voting system according to the course project description. The system consists of booths (implemented as individual processes) that can sentience with each other to maintain a total vote result consistently and robustly. [DownLoad or more ...] | | | | 
| Improve Web Search Relevance Using Users' Query Click-Through Log
MSN @ Microsoft, 06/2006~09/2006, Redmond, WA
I spent the summer of 2006 in Redmond, Microsoft's headquarter, working as an Research Software Engineering Intern in MSN Search. Overall it was a nice experience.
At that time, Ranking Relevance, i.e., the accuracy of ranking retrieved webpages according to users' searching goals, is still an important aspect that MSNers aimed to improve. Not surprising, my task was to research and find a feasible way that could improve the ranking relevance greatly in a short time. A mission impossible? [More...] | | | | 
| Build Relevance Ranking Function Using Boosting Algorithm
Web Search and Mining Group, Microsoft Research Asia, 06/2005-12/2005 With Lei Zhang, and Weiying Ma This project's primary purpose was to investigate the possibility to use advanced machine learning algorithms for web page relevance ranking.The motivation for using Boosting algorithm was the availability of a large dataset, with which more complicated features and sophisticated machine learning algorithms can be applied. Three aspects of the machine learning problems, i.e, feature, data, and learning algorithm, have been intensively studied. [More...] | | | | 
| Stereo Using Multi-flash Cameras
University of California, Santa Barbara, with Rogerio Feris
In this project we show how active illumination algorithms can produce a rich set of feature maps that are useful in dense 3D reconstruction. We start by showing how to compute a qualitative depth map from a single camera, which encodes object relative distances, being a useful prior for stereo. In a multiview setup, we show that along with depth edges, binocular half-occluded pixels can also be explicitly and reliably labeled. To demonstrate the usefulness of these feature maps, we show how they can be used in two different algorithms for dense stereo correspondence. Our experimental results show that our enhanced stereo algoritms are able to extract high quality, discontinuity preserving correspondence maps from scenes that are extremely challenging for conventional stereo methods. [More...] | | | |  | Face Annotation and Photo Album
Media Computing Group, Microsoft Research, Asia Automatic annotation of photographs is one of the most desirable needs in family photograph management systems. This project aims to build a tool for semi-automated face annotation in family photo album. The basic idea is to detect all the faces in a photo album and help users to annotate these detected faces for future retrieval. Based on the Bayesian similarity measure, which integrates face recognition and content-based image retrieval, the system is able to provide both face annotation and similar face retrieval, and thus significantly improve the management of family albums. [More...] | | | | 
| Face Detection and Recognition
Microsoft Research, Asia |
|
|