Xide Xia
Curriculum Vitae
52 Oxford Street Phone: (401) 209-4920
Cambridge, MA 02138 Email: [email protected]
EDUCATION
Harvard University, Institute for Applied Computational Science 2014 - Present
M.E. in Computational Science and Engineering
• Harvard IACS Student Scholarship 2015-2016 (awarded to two students per year in the department)
• GPA: 3.88/4.00
Coursework:
CS207: Systems Development for Computational Science CS182: Artificial Intelligence
AC209: Data Science AC298: Interdisciplinary Seminar in Computational Science
AM207: Monte Carlo Methods and Stochastic Optimization CS282: Reinforcement Learning
CS181: Machine Learning AC299: Independent Study and Research
CS205: Parallel Computing AC302: IACS Thesis Project
Brown University, School of Engineering 2012 - 2014
M.S. in Electrical Science and Computer Engineering
• GPA: 3.75/4.00
Coursework:
Pattern Recognition and Machine Learning Computer Vision
Scientific Programming in C++ Digital Image Processing
Mathematical Methods in Engineering and Physics I Mathematical Methods in Engineering and Physics II
Special Research Projects I Special Research Projects II
Beijing Institute of Technology, College of Information and Electronics 2008 - 2012
B.S. in Electrical and Information Engineering
• Renmin Scholarship (2008-2012, awarded to students with academic excellence and research potential)
• GPA: 86.7/100 Ranking: 9/73
PUBLICATION
2016
• Xide Xia, Mohammed AlQuraishi. “Computational prediction of protein-DNA interactions based on protein sequences information.” Forthcoming.
• Xide Xia, Finale Doshi-Velez, Pavlos Protopapas. “Cost-Sensitive Batch Mode Active learning: Designing Astronomical Observation by Optimizing Telescope
Time and Telescope Choice.” In In Proceedings of SIAM Data Mining Conference (SDM). 2016.
2015
• Mohammed AlQuraishi, Shengdong Tang, and Xide Xia. "An affinity-structure database of helix-turn-helix: DNA complexes with a universal coordinate
system." BMC bioinformatics 16.1 (2015): 1. (PMID: 26586237 Server: http://staging.proteindna.hms.harvard.edu/).
RESEARCH EXPERIENCE
Harvard Medical School, the Laboratory of Systems Pharmacology, Boston, MA June 2013 – Present
Graduate Research Fellow advised by Dr. Mohammed AlQuraishi and Professor Peter Sorger.
• Develop a new computational method for predicting protein-DNA interactions by implementing a whole machine learning pipeline for prediction protein-DNA
interactions.
• Build a model to simulate protein-DNA interactions, for which we formulate the problem as a convex optimization problem. In the model, we measure energies of
binding between protein and DNA molecules to derive the energies of interactions between atoms.
• Impose all the constraints and implement a regression problem that is specific to the formulation.
• Predict protein-DNA interactions based on protein sequence even if the structure information is missing (i.e. CIS-BP database).
• Implement large scale scientific computing in parallel and distributed environments.
• Assist Dr. AlQuraishi to construct an affinity-structure database in which structural and biochemical data of protein-DNA interactions are integrated:
http://staging.proteindna.hms.harvard.edu/.
Harvard University, Cambridge, MA
Intervention and Outcome Predictions: Understanding Vasopressor Administration and Weaning in the ICU Nov 2015 – Present
Research advised by Professor Finale Doshi-Velez.
• Analyze the MIMIC II 2.6 database gathered from four ICUs at the Beth Israel Deaconess Medical Center.
• Design a recurrent neural network model to simulate multidimensional physiological time series of patients before, during, and after vasopressor administration.
• Implement recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units in Torch-Autograd.
• Predict whether a patient will be administered a vasopressor.
Batch Mode Active Learning and Its Application to Astronomy Feb 2015 – Nov 2015
ME Thesis advised by Professor Finale Doshi-Velez and Professor Pavlos Protopapas.
• Developed a batch-mode cost-sensitive active learning approach that not only exploited uncertainty and representativeness of the whole unlabeled dataset, but also
took annotation cost into consideration.
• Designed a selectin criterion that combined uncertainty and representativeness by using a synthesized heuristic argument. Queried instances were selected from
different clusters to have a good representativeness of the whole unlabeled dataset. The selection criterion also considers the overall cost of the experiment while
keeping uncertainty and representativeness high.
• Applied the approach to optimize astronomical observations for object classification, including an extension that incorporates the possibility that nearby objects is
observed at the same time.
• Used two large astronomical data sets (MACHO and EROS) to demonstrate how the approach balanced the trade-off among FOV, aperture, and time cost.
• Results are published in In Proceedings of SIAM Data Mining Conference (SDM). 2016.
Texas Hold’em Poker Game AI Design March 2015 – May 2015
• Deterministic approach: with the probability of winning computed from permuting all the combinations, a fixed/deterministic strategy could be deduced.
• Human behavior adjustment: used hidden Markov model (HMM) to estimate opponent's hand strengths and applied Bayesian modeling with Markov chain Monte
Carlo (MCMC) to predict opponent's bluff tendency in each action.
• Texas hold’em poker game AI demo: https://sites.google.com/site/texasholdemdemo/.
Brown University, Providence, RI
Shape Recognition Feb 2013-May 2013
• Independent research advised by Professor Benjamin Kimia.
• Developed an accelerated algorithm in indexing into a large image database with graph editors.
• Built a simplification tree for shape searching based on k-nearest neighbors (KNN) algorithm.
Indoor Navigation System for the Visually Impaired Sept 2012 - Jan 2013
• Independent research advised by Professor Benjamin Kimia and Professor Pedro Felzenszwalb
• Designed a computer vision based guidance system for navigation in GPS-denied environments by locating user and finding out the optimal route to the destination.
North Carolina State University, Raleigh, NC June 2011- Sept 2011
• Summer Undergraduate research assistant at FREEDM lab advised by Professor Alex Huang.
• Built a network enabled battery monitoring system which solved the problem of battery managing in PHEV vehicle.
Beijing Institute of Technology, Beijing, China
Binocular Stereo Vision Camera Calibration Feb 2012- May 2012
• Honors Bachelor’s degree project and thesis advised by Professor Senlin Luo.
• Developed a stereo vision camera calibration system which estimated both intrinsic and extrinsic matrix of pin-hole camera models and corrected distorted images.
Multi-functional Artificial Intelligent Robot Mar 2011- Jun 2011
• National Undergraduate Independent Research Project advised by Professor Senlin Luo, which received the distinction award.
• Designed an obstacle-avoiding robot that records videos and sounds along its path.
Electronic Science & Technology Design Laboratory Sept 2009- May 2011
• Undergraduate research assistant for Professor Zhonglian Zhang.
• Assisted Professor Zhang’s research for the designing and the production of electronic devices such as intelligent vehicles, digitally controlled DC stabilized
voltage power supply, D/A converters, RF transceiver/receiver module, etc.
WORK EXPERIENCE
Harvard Medical School, Boston, MA June 2013 – Present
Graduate research fellow at Department of Systems Biology
Harvard University, Cambridge, MA Spring 2016 (Upcoming)
Teaching Fellow/ Grader: AM207 Monte Carlo Methods and Stochastic Optimization
Huawei Technologies Co., Ltd, Shenzhen, China June 2012- Aug 2012
Software development engineer intern
Agilent Technologies Co., Ltd, Beijing, China Sept 2011- Jan 2012
Undergraduate intern at mobile broadband division
North Carolina State University, Raleigh, NC June 2011 – Aug 2011
Summer undergraduate research assistant at FREEDM Lab
Beijing Institute of Technology, Beijing, China Sept 2009- May 2011
Undergraduate research assistant for Professor Zhonglian Zhang
PROGRAMMING SKILL
Python, C/C++, Java, Lua (Torch), HTML, JavaScript, R, Matlab, and Mathematica.
