Name: Kelvin Luu

I am a computer science major on my last year of school (4th year; undergraduate senior).

My main motivation for taking the course is to basically get more mastery over statistics - especially with respect to research. I have grown interested in the past year in topics in machine learning and natural language processing which heavily uses statistics. Furthermore, Bayesian probability is heaviliy seen in these types of sciences; e.g. what is the probability of A if I've seen evidence/data D? I think that 4780 gives me an opportunity to add another method of thinking or tool to my arsenal which may come in handy.

In particular, I took a course on topic modelling last semester that was extremely interesting to me. Topic modeling tries to classify sets of documents into some certain topics. More conceretely,S a topic can be seen as a set of words. So the set of words {government, president, united states} might be a topic about the government while {svm, error, emission} might be a topic on machine learning. The topic modelling problem is then two-fold: how to learn the topics and then the clustering of documents to each topic. Classical topic modeling uses heavy statistics to do so - frequently using probabilistic methods on co-occurence of words/terms.