Deep Unsupervised Learning. 3 credits, 3 contact hours

Catalog description

Prerequisites: DS 675 . The current era of data science requires identifying structure in high-dimensional signals. Transforming data into “good” representations can make the difference in the success or failure of downstream machine learning algorithms for classification, regression, or data generation. We will explore unsupervised learning algorithms that deal with linear and nonlinear structures and are also able to learn representations from multiple data modalities. We will use applications in computer vision, natural language processing, and audio processing to motivate our theory and algorithms.