Machine Learning Street Talk (MLST) #69 DR. THOMAS LUX - Interpolation of Sparse High-Dimensional Data
18 snips
Mar 12, 2022 
Dr. Thomas Lux, a research scientist at Meta in Silicon Valley, dives deep into the geometry behind machine learning. He discusses the unique advantages of neural networks over classical methods for high-dimensional data interpolation. Lux explains how neural networks excel at tasks like image recognition by effectively reducing dimensions and ignoring irrelevant data. He explores the challenges of placing basis functions and the importance of data density. Their ability to focus on crucial input regions reveals why they outperform traditional algorithms.
AI Snips
Chapters
Transcript
Episode notes
Delaunay Triangulation Explained
- Delaunay triangulation generalizes linear interpolation to higher dimensions by connecting data points to form simplices.
- These simplices create a piecewise linear interpolant, useful for approximating functions in higher dimensions.
Neural Networks vs. Classical Methods
- Classical machine learning methods like Delaunay triangulation partition space, but neural networks excel with high-dimensional, nonlinear data.
- Neural networks perform nonlinear dimension reduction, capturing important disparate sets in the input space.
Basis Function Placement in Neural Networks
- Neural networks place basis functions based on error, not just data density.
- This allows them to focus on areas where the underlying function changes most, unlike classical techniques.