Rushi’s Furnace

Kolmogorov Complexity and “Sophistication”

It seems to me that the most important discovery since Gödel was the discovery by Chaitin, Solomonoff and Kolmogorov of the concept called Algorithmic Probability which is a…

The Policy Gradient

The idea of policy-gradient methods is to change our policy using gradient descent methods such that our RL objective is maximised. Let’s see how.

The RL Ideology

The good thing about RL is that it possesses the power to produce exceptional results which can’t be made by a human. It’s the way to create an alternate intelligence which…

Bernoulli Restricted Boltzmann Machines

Generative modelling is the task of modelling the data distribution (which helps us generate new, similar, synthetic data points). There are multiple ways of dealing with…

An Expansive Probability Refresher

Many students study probability in high school or college but fail to grasp the common-sensical understanding of this natural topic. They usually end up memorising a ton of…