/reinforcement_learning_final_project

ML - Final Project - Product Recommendation in Online Advertising with Reinforcement Learning

Primary LanguagePythonMIT LicenseMIT

reinforcement_learning_final_project

ML - Final Project - Product Recommendation in Online Advertising with Reinforcement Learning

University of Colorado at Colorado Springs
PhD in Computer Science

Class: CS 4080-5080 - Reinforcement Learning - Fall 2021
Professor: Jugal Kalita
Student: Carlos Eugenio Lopes Pires Xavier Torres
E-mail: clopespi@uccs.edu
Date: October 20, 2021

Class Project
Product Recommendation in Online Advertising with Reinforcement Learning

Report

Carlos_Torres_Final_Paper.pdf

Presentation

Carlos_Torres_Final_Presentation.pptx

Carlos_Torres_Final_Presentation.mp4

Figures

  • Deep Q-Network

dqn

  • The RecoGym environment represented as a Markov chain of the organic and bandit user sessions. Adapted from RecoGym.

markov_chain_animated

  • System architecture.

system_architecture_animated

  • DQN internal architecture. Adapted from Nair et al. (2015).

dqn_architecture_animated