- Used CorEx to learn a representations of complex data based on optimizing an information-theoretic objective
- Explained the correlations in the data as measured by multivariate mutual information.
- Explored CorExs efficiency and correctness from two aspects: clustering and predicting
Discovering Structure in High-Dimensional Data
- Proposed the concept of subjectiveness and objectiveness of points and links in RDF resource graph
- Modified the restart probability of original Page-Rank algorithm to concentrate more on user’s query
- Completed iterations between points and links to achieve points’ final importance and ranked resources
Query Dependent Ranking on RDF Graph, Tsinghua University
- Based on model-view-control architecture pattern, developed blog web application with Python under webapp2 framework
- Incorporated cache technology to relief pressure of database by using google app engine api memcache
- Considered user information security by using hmac hash
Blog Web Application
- Build a polar scientific data search engine based on Apache Solr and Apache Nutch.
- Developed deduplication algorithm and built plugin for Nutch.Used Nutch to crawl three polar data sets.
- Built relevancy graph for crawled data based on Jaccard Similarity with meta information from Tika.
- Designed both content-based and link-based ranking algorithm, combined these algorithms to rank the result of search based on Solr.
Polar Scientific Data Set Search Engine
- Predicted click possibilities of advertisements given Avazu Dataset
- Used random forest to rank and select most significant features from high dimensional data set.
- Built SVM, Nave Bayes, Perception, Logistic Regression models, and compared prediction results.
- Combined above models and build a new blending model using neural network to achieve better prediction.
Click-Through Rate Prediction on Avazu Dataset
- Designed mobile application on Android to achieve long-distance smart control using different protocols.
- Implemented server with logical ability and simulated different sensors through JAVA programming.
- Completed communication among different parts of the system using WIFI, UDP and HTTP protocols.
Android based Home Service System
- Helped make new market strategy based on China Unicom customers’ data analysis result.
- Extracted data from JSON format files and finished data cleaning and preprocessing with JAVA.
- Clustered users with 3G data traffic information using K-Means algorithm based on Hadoop frame.
- Found frequent pattern with pruning included FP-Growth algorithm and made new market strategy.
China Union Customers Data Analysis, China Union
Hello, my name is Xinyue Zhou! Nice to meet you!
Here is little about me......
I am currently a master student in University of Southern California. My major is Computer Science. And I love this area very much!
I have more than 4 years experiences related to software engineering. These experiences include Android application, web application, data analytic, machine learning, database, search engine...... All of these valuable experiences grant me strong skills in programming and also high ability in creative thinking.
Besides technology, I am a people enjoy cooking and traveling. So, if you cannot find me in front of computer, I must be in kitchen or on the road!