Average Reward, New Problem Setting: Continuing Tasks and How to Avoid Discounting

We got used to using discounting for almost every single task so far. Well there are some problems that we realize what we do makes no sense. Here I am introducing Average Reward, how is it better to use, and a problem setting where it is the only option so far: Continuing tasks. And afterwards we will be seeing the first algorithm that uses average reward: Differential Semi Gradient SARSA

6 min read

Semi-Gradient Control Methods

I will be going over how Semi-Gradient SARSA works, first theory and then raw linear python implementation. I love learning things in the simplest form possible, without a lot of fancy stuff hidden in between. Well, if you are like me I believe you will enjoy this series for BetterRL.

8 min read

Counting Sort

Counting sort is a nice in-place sorting algorithm that we can use for sorting instantly. (This is mostly used for competitive programming.) What I meant by this is, we can use counting sort when we are getting the input. This will not get any additional cost for us, and really good technique for using in-place sorting. I’ll get there after explaining the algorithm.

3 min read

Factorizing a Number

Before starting- What is Prime Factorization ? What is a Prime number ? If you are curious about these please checkout this link before proceed because I will not explain them here :)

5 min read

Kadane’s Algorithm

What we will learn today is, how to find the sum of maximum of the subarrays, in optimal solution. First, let’s clarify the goal a bit.

3 min read

Machine Learning Basics

Reaaally introductory post for Machine Learning basics. What is Supervised and Unsupervised learning?

2 min read