Deep Learning and the Relational Algebra

Zack Ives (University of Pennsylvania) https://simons.berkeley.edu/talks/zack-ives-university-pennsylvania-2023-11-15 Logic and Algebra for Query Evaluation Increasingly, machine learning inference is playing a key role in all aspects of query processing. We can train a model over a view over a database, or make predictions over a data stream, or group data by predicted class. Do we get any benefit by extending the relational algebra with tensor operations, autograd, and other primitives in deep learning dataflows? We relate our early, work-in-progress experiences in tackling this question.