∫ntegrabℓε ∂iﬀerentiαℓs · unorganised notes, code, and writings of random topics

Gaussian copula and multivariate normal distribution

January 24, 2019 blog code math

This is a failed attempt to simplify the process to generate multivariate Gaussian distribution by utilizing the copula function. However, it has the merit of figuring out what information a copula function failed to provide. [more]

Agrawal et al (2018) A rewriting system for convex optimization problems

January 18, 2019 paper, code

I learn about this paper when I am studying the python package cvxpy. This is an interface to other solvers such as glpk and Boyd who authored the book of Boyd & Vandenberghe is one of the creator. This package, however, requires some special formulation of the problem called “disciplined... [more]

Wang, Youssef, & Elhakeem (2006) On some feature selection strategies for spam filter design

January 7, 2019 paper

A short paper on comparing different algorithms to do feature selection in spam filtering problems. [more]

Lin, Tegmark, & Rolnick (2017) Why does deep and cheap learning work so well?

January 2, 2019 paper

The paper points out in sec 1 that, in contrast to the good old-fashioned AI (GOFAI) algorithms, artificial neural networks are understood only at a heuristic level. The authors try to provide analytic insights on why deep learning and general ANN works. [more]

Das & Stein (2011) Differences in Tranching Methods

December 31, 2018 paper

There are two methods the rating agencies used to evaluate credit worthiness. S&P and Moody’s in particular, uses the probability of default (PD) and the expected loss (EL) respectively. This paper compares these two methods in the context of CDO. [more]