|Asan engineering student, I am applying for your graduate program in computerscience to reinforce my computer science expertise and seek a career change incomputer science, especially in machine learning and data mining. In the pastyears, along with my growing academic and professional experiences relevant tocomputer science, I also witnessed the magic that computer science creates tomake our world a better one. At this point of time, right in the era of bigdata and cloud-computing, I would like to systematically study computer scienceand be a part of this splendid innovation.
Peoplechange, so do their interests. Hence it is no wonder that with my undergraduatemajor in materials science and graduate specialty in automotive engineering, Iam doing the transition, again. However, when I look back and re-examine myprevious experiences, actually I find that everything is connected and natural,all these experiences have completed me and led me to the new challenges. Forexample, when I was at Beihang University as an undergraduate, my studies inmaterials science consolidated my analytical skills and acumen towards technologies.Without them, I would not have been recommended by the school for the directadmission to Polytechnic University of Turin where my interest in computerscience originated.
Comparedto my previous experimental study in materials science, learning automotiveengineering in Italy offers me a distinctive perspective to focus oninformation science and logic. Early this year I took part in a sponsoredproject on unsupervised real-time seed detection for ROJ Company, an Italiancompany focusing on advanced electronics design in mechatronic and embeddedsystems. This project honed my programming skills by using Python and pushed meto relearn C programming. I really enjoyed the achievement gained by employingvarious techniques to make the detection more accurate. I experimented bothmean shift and k-means algorithm to segment images and find the “bulk” patternof object. I taught myself Machine Learning and Algorithms from Coursera, aimingto strengthen my theoretical foundation. Also, I read numerous papers in objectclassification and tested neural network and support vector machine (SVM)respectively to complete object classification task. Among the plenty methodsread from papers, I realized Bag-of-Features model could be tailored to ourinterest and I used Speeded Up Robust Features (SURF) to describe objectfeatures and SVM as a classifier. Up to then, the accuracy had reached 89.51%.Challenging yet rewarding, the accumulated experience dealing with these imagesgave me more insights to solve it. I came up with another unsupervised modelwhich completed the classification task based on statistical information andimproved the accuracy to 97.20% with nearly 20% decrease in algorithm runningtime.
Inthe fall semester, I went back to Beihang University and joined the IntelligentComputing and Machine Learning Lab. I have been mainly focused on semanticparsing – a natural language processing (NLP) field. After familiarized with fundamentalNLP topics like sentiment analysis and semantic parsing by teaching myselfonline NLP course, 数减edted by others ously fascinated...st straight A and ranked asone of the tops in my class. to be and focusI also packed in mItested several question answer models including geography and travel query byusing Java. Recently I began to build an interactive learning model to learnthe conceptual space of software agents. I have already completed the Grammar part which employed dynamicprogramming to return all possible parsers. Currently, I am working on the Feature part which automatically extractsmatching relationships. Later I would use a stochastic gradient descent methodto train this model. At the same time, in order to reinforce my foundations, Iattended Advanced Algorithm, Statistical Learning and Database Principlecourses whose credits are about to obtain. The research experience consolidatesmy programming skills and give me a deeper understanding of NLP field.
Apartfrom academic research at Beihang, I have simultaneously been interning with Huawei, one of themost influential information and telecommunication manufacturers in the world.As a data mining intern, I am fully involved in a user behavior research intelecommunication industry. I collaborated with other team members to designalgorithm and analyze the user group with high customer churn rate. I extractedand cleaned extensive data of user online browsing behavior and historicalconsuming data, combined them with their personal information to form userfeatures. Then I implemented decision tree C4.5 algorithm to predict the mostlikely customers to lose. With carefully selected variables and adjustment oftree structure, the results revealed the features of lost customers. I showedthe visualized data by using Python and Tableau to internal business team andcollaborated with them together to come up with corresponding strategies, whichwas potentially possible to save more than 20% customers.
Atthis point of time, I almost see my insufficiency of computer science as thebarrier on my career transition, although I have spared no effort to complementmy skills. All of my previous experiences are closely related to data miningand machine learning, but to break the barrier, I need systematic trainings in computerscience theories and specific courses in algorithm, programming implementation,software engineering and others.
Uponcareful consideration, I choose the specifically designed Master of Science inComputer Science program for Scientists and Engineers offered by University ofSouthern California to base my graduate studies. By taking advantage of thecomprehensive two-year program, not only can I consolidate my computer sciencefoundation with a coverage in programming, operating systems, algorithms andothers, but I can also get involved in internships and research opportunities.For instance, I expect to have the privilege taking part in the directedresearch with Dr. Fei Sha in the Statistical Machine Learning Lab. Meanwhile, Ifound the ICT Natural Language Dialogue Group an excellent match with myprevious NLP research at Beihang University. Besides, the featured IndustrialAffiliates Program with an array of high-profile IT enterprise members at USCdistinguishes the computer science program apart from others. With my burningpassion and accumulated experience in computer science, I sincerely lookforward to be accepted by your rigorous and supportive program.