Jan Malte Lichtenberg

I blog about reinforcement learning on

[Jan Malte Lichtenberg @JanMalteL]

Understand how #ReinforcementLearning pseudo code translates to learning progress, one line at a time, in my new interactive blog series https://t.co/vRATGWnYUn The first post is about Q-learning. Inspired by @distillpub and @ch402. Made with #p5js.

[Jan Malte Lichtenberg @JanMalteL]

Maybe a good day to re-post my tutorial on *Q-learning*. Learn about the algorithm one line at a time. :)

[Jan Malte Lichtenberg @JanMalteL]

Understand how #ReinforcementLearning pseudo code translates to learning progress, one line at a time, in my new interactive blog series https://t.co/vRATGWnYUn The first post is about Q-learning. Inspired by @distillpub and @ch402. Made with #p5js.

[Jan Malte Lichtenberg @JanMalteL]

Understand how #ReinforcementLearning pseudo code translates to learning progress, one line at a time, in my new interactive blog series https://t.co/vRATGWnYUn The first post is about Q-learning. Inspired by @distillpub and @ch402. Made with #p5js.

[Jan Malte Lichtenberg @JanMalteL]

@_philschmid This is pure gold! Thanks Philipp!

[Jan Malte Lichtenberg @JanMalteL]

#genuary2022 - Day 12 - Query: "Packing (squares, circles, any shape…)"

https://t.co/xofpgiEyUp #genuary #p5js

[Jan Malte Lichtenberg @JanMalteL]

#genuary2022 - Day 10: Machine learning, wrong answers only.

Finding the data that explains the model.

#genuary #p5js

[Jan Malte Lichtenberg @JanMalteL]

The post also contains an intuitive comparison between Sarsa and Q-learning (or more generally, "on-policy" vs. "off-policy" learning) and describes how both approaches result in different behavioral policies. I hope you enjoy reading it. #Pancakes

[Jan Malte Lichtenberg @JanMalteL]

Learn about the SARSA reinforcement learning algorithm and its differences to the Q-learning algorithm: https://t.co/HXLBaDdmsw

This is the 4th post of my interactive blog series "RL: line by line".

#RL #p5js #tutorial

[Jan Malte Lichtenberg @JanMalteL]

Track the learning progress of your agent and play around with different learning parameters such as the exploration parameter or the learning rate.