Reinforcement Learning for Food Waste Reduction within Albert Heijn

on YouTube ↗32m

" summary

Applying reinforcement learning to reduce supermarket food waste by setting increasing markdowns on near-expiry products, balancing markdown costs against items destroyed. The approach models the problem with a discrete state and action space, learns policies from historical data using offline Q-learning, uses off-policy evaluation to assess new policies (with discussion of their limitations), and validates performance via production experiments and an implementation-oriented setup.

AI-written summary · about

" chapters (13)

1. 0:00Food waste at Albert Heijn
2. 1:04Global food waste
3. 2:30Food supply chain
4. 3:55Dynamic markdown
5. 6:21Rule-based markdown
6. 8:09Why reinforcement learning
7. 10:14Simple RL model
8. 12:10Offline learning
9. 13:05Performance evaluation
10. 16:23Off-policy evaluation
11. 20:12A/B tests in stores
12. 22:05Conclusions
13. 23:53Q&A

" related talks

• 49mDon’t throw away food! AI and Data to reduce food wasteSantiago Salgado, Javier R. Azpeitia · T3chFest 2023
• 41mBetter Learning from the Past: Counterfactual / Batch RLEmma Brunskill · Emerging Challenges in Deep Learning (Simons Institute workshop, 2019)
• 47mAdvanced Reinforcement Learning & Data Science for Social ImpactVukosi Marivate · Deep Learning Indaba 2018
• 34mReinforcement Learning in Business ApplicationsAnand Oka · Global Artificial Intelligence Virtual Conference Sep 2020
• 28mHow Human Expertise and AI drive Scalable Forecasting Success atRuben van de Geer, Anant Pingle · Devworld Conference 2025
• 34mOffline Reinforcement Learning and Model-Based OptimizationSergey Levine · Reinforcement Learning from Batch Data and Simulation
• 26mHacking Reinforcement LearningGuillem Duran · PyConES 2018
• 27mUsing Reinforcement Learning to Optimize IAP Offer Recommendations in Mobile GamesEmiliano Castro, Vinicius Alves, Felipe Antunes · Ray Summit 2021