Lab1: A noisy-channel model for spelling correction. This means creating an edit model (likelihood terms) and a language model (distribution in the noise channel model). At test time, a sentence with exactly one error will be passed into your edit model which will perform various edits and select the correction that gives the highest likelihood under the language model.
Lab2: Part-of-Speech tagging for English, Bulgarian and Japanese. Viterbi algorithm is used during analysis.
Lab3: Answering questions regarding to Proabilistic Context Free Grammar(PCFG). No coding involved.
Lab4: Sentiment Analysis on IMDB movie reviews. Implenmented Naive Bayes classifier to accomplish this task.