Explore deep symbolic regression for recurrent sequences in this comprehensive video featuring an interview with first author Stéphane d'Ascoli. Delve into the innovative use of transformers for symbolic computation on integer and floating point number sequences. Learn about the clever encoding of input space, data generation process, and the model's ability to predict and represent sequences from the Online Encyclopedia of Integer Sequences (OEIS). Discover the potential applications beyond number sequences, success and failure cases, and experimental results. Gain insights into overcoming research challenges and interact with the provided demo to experience the model's capabilities firsthand.
Predicting the Rules Behind - Deep Symbolic Regression for Recurrent Sequences
Overview
Syllabus
- Introduction
- Summary of the Paper
- Start of Interview
- Why this research direction?
- Overview of the method
- Embedding space of input tokens
- Data generation process
- Why are transformers useful here?
- Beyond number sequences, where is this useful?
- Success cases and failure cases
- Experimental Results
- How did you overcome difficulties?
- Interactive demo
Taught by
Yannic Kilcher
