Talks

Back Propagation Through Time (BPTT) for simple RNN

• If you are familiar with backpropagation for simple fully connected neural networks using computational graphs, then you can jump to 1:27.

Application of Singular Value Decomposition (SVD) to Image compression

• In this video, I show you have SVD could be applied to images (which are essentially a matrix of elements)
• We can use rank-k approximation to represent an image which eventually leads to image compression

Vector Space: Independence, Span, Projections (in Tamil)

• If you are in a hurry to conceptually understand the core Linear algebra concepts, then this video is for you.
• Prerequisites: Matrix, Vector, matrix-vector product

Approximation is all you need: Fourier Transform

• Engineers only do approximations!
• In this video, I introduce you to one of the most profound mathematical transformations called the Fourier Transform.
• If you are too rigid about definitions, then in this video I talk about “The Fourier series”.