Mathematician & Educator
Self-contained public lectures, keynote addresses, and research seminars.
This video provides an intuitive and visually driven introduction to the Fourier Transform, explaining its fundamental principles by decomposing complex signals into constituent sine and cosine waves. It demonstrates how this transformation maps a function from its original domain to a frequency domain.
Derives Claude Shannon's 1948 entropy formula H(X) = -Σ p(x) log₂ p(x) from Kraft's inequality on binary prefix trees and connects lossless compression directly to language modeling.
Solves a formidable combinatorics problem over subset sums divisible by 5 by encoding subsets into polynomial products and filtering coefficients with complex 5th roots of unity.
Bridges PDE physics and pure harmonic analysis: shows why sinusoidal eigenfunctions naturally diagonalize the heat operator and how inner products isolate each rotating frequency.
Geometric dissection of token embeddings, unembedding matrices, high-dimensional almost-orthogonality (Johnson-Lindenstrauss intuition), and Transformer residual streams.
Explains why every LLM minimizes cross-entropy loss: measuring the exact bit-rate penalty D_KL(P || Q) incurred when compressing true distribution P using predicted distribution Q.
Search or browse every recorded session across courses and talks.
