Statistical Foundations of Natural Language Processing

Treats language as data: you build probabilistic models that turn raw text into something a machine can count, classify, and predict from, starting with tokenization and n-grams and working up through HMMs and vector-space embeddings. Most of the grade comes from three programming projects and a term project where you implement a real NLP pipeline end-to-end, with a final exam covering the statistical theory behind it. It sits at the intersection of EE signal-processing intuition and modern ML, and gives you the foundations you'd need before tackling neural language models, computational social science work, or any LLM-adjacent research.