Spline Theory Insights
Randall discusses the revolutionary spline theory of neural networks, highlighting how deep networks can be understood as continuous piecewise affine splines. He explains that the input space is divided into convex regions where the network behaves as an affine mapping, shedding light on the network's learning process and its ability to extrapolate. The analogy of a honeycomb structure illustrates how decisions are made within these regions.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Want to Understand Neural Networks? Think Elastic Origami! - Prof. Randall Balestriero
Related Questions