Machine Learning

Machine learning (ML) is sometimes treated as synonymous with “artificial intelligence.” I myself wouldn’t go that far—not unless you are willing to conceptualize OLS regression as a form of artificial intelligence (we’ll talk more about that in the Nov. 6 meeting). A lot of what we’ve already talked about—especially topic modeling and word embeddings—is classified under the heading of unsupervised machine learning.

That said, ML is certainly a critical component of the modern (generative) AI landscape. You’ll see that more clearly in the Nov. 13 class when we talk about the Transformer architecture—the bedrock for cutting-edge text classification and large language models. Most of our discussions over the next couple weeks will deal specifically with supervised machine learning—or using labeled data to train models to predict unlabeled data.