What do exploring an unknown labyrinth and extracting a #MachineLearning model have in common? ➡️Both are problems of exploration and mapping in an unknown environment.

All algorithms (>350) are open-source at https://t.co/J3sIcBxqAs under an open-data license. If you enjoy coding and optimization, playing TIG is a great way to learn, like a permanent hackathon with a dynamic team and community. Happy optimization, everyone!🎅🚀

You can track the live usage of all algorithms here: https://t.co/aIwnu49yUp. It's fascinating how their use and main parameters evolve as benchmarkers compete with them to mine qualifying solutions...

A safe Rust implementation of HGS for the VRPTW is now live and running on The Innovation Game! Rewriting it in $Rust forced me to rethink many design choices, leading to a code base of around 2000 lines that is simple, modular, and SOTA for #VehicleRouting #Optimization #TIG

Many thanks to Awa Khouna and Julien Ferry for this excellent work. Link to the paper and code below: https://t.co/3u2Ma1OpG9 https://t.co/oCHRefYQ10 See you soon in San Diego!

This #OnlineOptimization work helps formalize how explainability can inadvertently expose models to reconstruction, and quantifies the number of queries required by an attacker... an essential step toward safer and privacy-preserving #MachineLearning

The simple algorithm he proposes can perfectly reconstruct decision trees and random forests from prediction and explanation queries, with provable efficiency bounds rooted in online competitive analysis.

In his #neurips2025 paper "From Counterfactuals to Trees: Competitive Analysis of Model Extraction Attacks", our talented PhD student @AwaKhouna turns this analogy into a rigorous mathematical framework, building bridges between online exploration and model extraction attacks.

Propulsing vehicle routing into space... At @IVADO_Qc Digital Futures Event, my brilliant postdoc Théo Guyard will show how #Optimization & #MachineLearning meet orbital dynamics to clean up space debris, in collaboration with @NASA Ames Research Center.🚀 https://t.co/lNGpfFR81P

OCEAN is evolving 🌊, check it out! https://t.co/KJInvrKxBZ v2.0 offers a one-click Python library for optimal counterfactual explanations in tree ensembles (RFs, boosting...) based on MILP and CP models. Install with "pip install oceanpy". Unlike unstable heuristics that may

A thought-provoking model of acceptance decisions in academic journals & conferences: slashing rates doesn't cut the eventual number of accepted papers, but it does delay good work, inflate visible submissions, and pile up reviewing/revision load:

Another interesting VRP challenge to keep an eye on 👀

The CVRPLIB introduces a new XL benchmark set (1,000–10,000 clients) along with a BKS challenge in January 2026. 30 days. One real-time leaderboard to compare modern methods: from classic OR and ML approaches, to learning-augmented hybrids and decomposition strategies...

#ORMS #Analytics quietly powers decisions in healthcare, supply chains & beyond... yet is largely invisible to the public. Drawing inspiration from the growth of #ML, we outline 10 key actions to help move this status quo: https://t.co/kXcwwFsQzS. What else would you suggest?

Maria has been the mastermind behind this innovation, from the original idea to the full methodology and implementation... an impressive achievement, especially for a first PhD paper! Congratulations !!

The awarded paper, "Solving Two-Stage Programs with Endogenous Uncertainty via Random Variable Transformation" ( https://t.co/kc5KLsXwwt), introduces a general transformation technique that opens new pathways for solving two-stage programs with #EndogenousUncertainty.

Maria Bazotte (@MBazotte), one of our brilliant PhD students at the SCALE-AI Chair at @polymtl, has just been awarded the Dupačová-Prékopa Best Student Paper Prize in #StochasticProgramming at #ICSP2025!

Work done at the SCALE-AI Chair at @polymtl, with my fabulous co-authors Arthur Ferraz, Quentin Cappart, Axel Parmentier (@ParmentierAxel1 ), Alexandre Forel, and Cheikh Ahmed... Happy #ORMS, #StrategicOptimization, and #MachineLearning, everyone!

Related links: 🔗Paper 1: https://t.co/8SvqXtWIQX 🔗Paper 2 (NeurIPS 2024): https://t.co/6r5ftyh1rn 🔗Source code:

How can GNNs be harnessed for an efficient strategic optimization of delivery districts using ML+OR pipelines and end-to-end learning? Check @dataskeptic's latest podcast discussing 2 of our recent works on this topic (my intervention starts around 5:00):