Using Reinforcement Learning to Optimize IAP Offer Recommendations in Mobile Games
summary
Uses offline reinforcement learning with behavior cloning then policy improvement, running on a Kubernetes cluster with Ray (RLlib and Tune) to recommend IAP offers in Sniper 3D. Trains from logged play traces, deploys a cloned policy, collects incremental batches, and fine-tunes with an offline algorithm (MURAL-style) to avoid catastrophic forgetting. Results show a 3% LTV uplift versus the supervised baseline, 90% faster response time, and 94% lower training cost during the proof of concept.
AI-written summary · about
chapters (7)
- 0:00Introduction to RL for IAP
- 6:37Problem framing as RL
- 8:47Building the RL agent
- 15:06Learning strategy and results
- 20:43Key performance metrics
- 24:18Next steps and future plans
- 26:26Conclusion and Q&A
related talks
30mReinforcement Learning - ChatGPT, Playing Games & More
12mPredictive Response Optimization: Using Reinforcement Learning to...
01h 17mTraining intelligent game agents using deep reinforcement learning
25mDon't do it like a virgin – Automated revenue management in gaming
22mAWARE: Automate Workload Autoscaling with Reinforcement Learning in Production...
31mDesigning game AI with reinforcement learning
27mEnhancing Personalization Using Recommendation Systems & Multi-Armed Bandits
14mConnecting Reinforcement Learning to Simulation Software