Explore the intersection of deep learning and symbolic mathematics in this 55-minute video from Launchpad. Delve into neural networks and symbolic reasoning, learn about generating random expressions, and understand formulae for counting trees and expressions. Discover techniques for creating datasets for integration and ordinary differential equations, followed by dataset cleaning and statistics. Examine sequence decoding, evaluation methods, and comparisons with mathematical frameworks. Investigate generalization across and beyond generators, gaining insights into the potential of deep learning in symbolic mathematical problem-solving.
Overview
Syllabus
Intro
Neural networks and symbolic reasoning
Problem statement
Generating expressions randomly
Counting number of expressions
Formulae for counting no. of trees and expressions
Generating dataset for Integration
Ordinary Differential Equations (1st order)
Dataset Cleaning
Dataset Statistics
Sequence Decoding
Evaluation
Comparison with mathematical frameworks
Generalization across generators
Generalization beyond the generator
Conclusion
Taught by
Launchpad
