Method
Wide & Deep
Joint linear memorization and deep generalization model.
9 cases from 6 companies apply Wide & Deep. 5 were checked in an online A/B test; 4 report results in production.
Typical tasks: Recommend to new users and new items, Retrieve and rank from millions of items, Evaluate models, Optimize decisions and 6 more. Companies: Instacart, LinkedIn, eBay, Stripe, DoorDash and 1 more. Often combined with: Feature selection, A/B test, Threshold selection, Propensity / response modeling, Auctions and bidding and 13 more.
| Empowering Carrot Ads with Domain Adaptive Learning | adtech | 2026-05 | |
Traffic Filtering in Programmatic Ads![]() | ![]() | adtech | 2024-05 |
| One model to serve them all | adtech | 2023-12 | |
| Evolving Recommendations: A Personalized User-Based Ranking Model | e-commerce | 2023-10 | |
| How we built it: Stripe Radar | fintech | 2023-03 | |
| Homepage Recommendation with Exploitation and Exploration | food-delivery | 2022-10 | |
| Calibrating CTR Prediction with Transfer Learning in Instacart Ads | adtech | 2022-09 | |
| Challenges and practical lessons from building a deep-learning-based ads CTR prediction model | adtech | 2022-08 | |
| Matching external companies to LinkedIn's Economic Graph at scale | social | 2022-04 |
