Mar 26, Mon
Damage Control: The Essential Lessons of Crisis Management (Chapter 1-6)
In light of my coming internship at Uber ATG, and the recent PR crisis faced by Uber and Facebook, I find it very amusing and surprisingly relevant to read this book. Mostly read on the plane SEA -> DTW, a few simple things I learnt:
- Crisis needs leadership that is already strong.
- Crisis needs good and quick actions, not perfect ones.
- Figure out who / what you are playing against.
Mar 27, Tue
MLAPP (Chapter 18: Kalman Filter)
Brushing up my Gaussian conditioning / marginalizing skills and essentially appy Bayes rules on Linear Gaussian Systems (Linear tranisiton, Gaussian observation model).
Mar 29, Thu
Original paper of A* Search
A Formal Basis for the Heuristic Determination of Minimum Cost Paths.
I read the paper mostly for the proofs related to A* search and find it pretty amazing that heuristic search can actually yield the optimal solution under very mild conditions. The proofs are not very easy to come up with all by myself, I gave up after trying all afternoon and refer to original paper.
Udacity FlyingCar: Motion Planning
Mar 30, Fri
Resnet original paper
Our study group worked on understanding how the shortcut is made while changing block size (they use padding and 1x1 convolution with stride). We also found out the convolution can actually be matrix multiplication not just elementwise multiplication... so they map to more or less channels.
Mar 31, Sat
MLAPP(4.1 & 4.4) Multivariate Gaussian
Schur Complement: block diagonalization.
Matrix Inversion Lemma (rank one update).
Linear Gaussian System Bayes rule (to be used in Kalman Filter).