Recommender Systems Research
This course focuses on theoretical foundations and practical applications of deep learning, the subfield of machine learning concerned with large neural networks trained on large data sets. Topics include training models by stochastic gradient descent, implementing various neural network architectures, and choosing network hyperparameters. Application areas include classification, regression, and reinforcement learning problems. Students will implement their own neural networks from scratch and get experience using state-of-the-art deep learning libraries.
* CSC 221
* MAT 140 or MAT 150 or MAT 160
The ubiquitous ".. people who viewed this item also viewed these items .. " recommendations found in online shopping applications are driven by underlying Recommender Systems. Started in the 1990's, these systems have evolved from relatively simple news recommenders to today's sophisticated recommender systems used by Amazon and Pandora. This course begins with an overview of the motivation for recommenders and the fundamental recommender methodologies: collaborative filtering, content-based, and hybrid. Then, we will do a deeper dive into the popular collaborative filtering (CF) algorithms: user-based, item-based, and matrix factorization. We will use Python programming and research datasets containing movie ratings to explore how these CF algorithms produce recommendations. Finally, we will learn how to evaluate recommender results with established research metrics and protocols as well as propose and evaluate custom modifications to recommender algorithms.
Satisfies Applications elective in the Computer Science major and minor.
Counts an an elective in the Data Science inerdisciplinary minor.
CSC 221, MAT 150, and the ability to program in a high-level language such as Python, Java, or C++ at the level expected in CSC 221.
|Time|| T R
Time 1135 - 1250pm