Abu Dhabi Machine Learning Season 2 Episode 5

 05.09.2022 -  Abu Dhabi Machine Learning -  ~2 Minutes


  • Monday, September 5, 2022 from 5:00 PM to 6:00 PM (Abu Dhabi Time)


  • At your home, on zoom. We are looking at hosting in-person events (and broadcasting online on zoom). Any companies or institutions interested in hosting us?

The Meetup page of the event:

Abu Dhabi Machine Learning Meetup Season 2 Episode 5


Talk 1: Optimizing Supply Chain with Reinforcement Learning and Graph NNs

Abstract: Distribution networks are a crucial part of supply chains that often entail highly complex optimization problems. For example, optimizing the transportation cost of multiple kinds of goods over a time horizon with order consolidation requirements is an NP-hard problem. In this presentation, we will explore a distribution problem with such characteristics, called the Shipping Point Assignment (SPA) problem. Inspired by recent advances in combinatorial optimization using reinforcement learning and graph neural networks, we propose a deep Q learning agent with a GNN-based Value Function Approximation to optimize the SPA problem. We compare this agent with other RL-based and deterministic approaches in order to determine whether distribution networks of this kind could be better solved by ML-based techniques.

Speaker: Javier Porras-Valenzuela is a Computer Scientist based in Costa Rica currently working as a Machine Learning Engineer at Encora. He has a Master’s degree in Computer Science from Instituto Tecnológico de Costa Rica (ITCR). He has over seven years of experience in software, six of them as a Machine Learning Engineer working in domains such as supply chain, finance and ad-tech. His areas of interest include Operations Research, Reinforcement Learning, Graph Neural Networks and Natural Language Processing.


Video Recording of the ADML Meetup on YouTube

  • YouTube videos:

ADML S2E5 - Optimizing Supply Chain with Reinforcement Learning and Graph NNs

ADML S2E5 - Optimizing Supply Chain with Reinforcement Learning and Graph NNs