Winning AI4TSP: Solving the Travelling Salesperson Problem with Self-programming Machines - Shopify

This post details Shopify's participation and first-place win in both tracks of the AI4TSP Competition. For Track 1, the team used dialectic search to find a tour that performed well on average across various travel time scenarios. For Track 2, deep reinforcement learning was employed to build the tour dynamically. The post emphasizes the concept of 'self-programming machines' where algorithms learn and adapt, highlighting the success of this approach in solving computationally hard problems with uncertainty.