# Warmer for Less: A Cost-Efficient Strategy for Cold-Start Recommendations at Pinterest

arXiv: https://arxiv.org/abs/2512.17277 · HTML: https://mlsanity.com/paper/2512.17277/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 71 references that are in the map.
- First posted on arXiv 19 Dec 2025.

## Builds on (most cited first, 30 of 30)

- [Deeper, Broader and Artier Domain Generalization](https://mlsanity.com/paper/1710.03077/) (2017)
- [Learning to Generalize: Meta-Learning for Domain Generalization](https://mlsanity.com/paper/1710.03463/) (2017)
- [Domain Generalization via Invariant Feature Representation](https://mlsanity.com/paper/1301.2115/) (2013)
- [Domain Generalization via Model-Agnostic Learning of Semantic Features](https://mlsanity.com/paper/1910.13580/) (2019)
- [DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems](https://mlsanity.com/paper/2008.13535/) (2020)
- [Unified Deep Supervised Domain Adaptation and Generalization](https://mlsanity.com/paper/1709.10190/) (2017)
- [Deep Domain-Adversarial Image Generation for Domain Generalisation](https://mlsanity.com/paper/2003.06054/) (2020)
- [Unbiased Learning-to-Rank with Biased Feedback](https://mlsanity.com/paper/1608.04468/) (2016)
- [Learning from Extrinsic and Intrinsic Supervisions for Domain Generalization](https://mlsanity.com/paper/2007.09316/) (2020)
- [Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System](https://mlsanity.com/paper/2010.15363/) (2020)
- [Managing Popularity Bias in Recommender Systems with Personalized Re-ranking](https://mlsanity.com/paper/1901.07555/) (2019)
- [Contrastive Learning for Cold-Start Recommendation](https://mlsanity.com/paper/2107.05315/) (2021)
- [Self-Supervised Hypergraph Convolutional Networks for Session-based Recommendation](https://mlsanity.com/paper/2012.06852/) (2020)
- [Warm Up Cold-start Advertisements: Improving CTR Predictions via Learning to Learn ID Embeddings](https://mlsanity.com/paper/1904.11547/) (2019)
- [Contrastive Self-supervised Sequential Recommendation with Robust Augmentation](https://mlsanity.com/paper/2108.06479/) (2021)
- [Deconfounded Recommendation for Alleviating Bias Amplification](https://mlsanity.com/paper/2105.10648/) (2021)
- [Learning to Warm Up Cold Item Embeddings for Cold-start Recommendation with Meta Scaling and Shifting Networks](https://mlsanity.com/paper/2105.04790/) (2021)
- [TransAct: Transformer-based Realtime User Action Model for Recommendation at Pinterest](https://mlsanity.com/paper/2306.00248/) (2023)
- [ESAM: Discriminative Domain Adaptation with Non-Displayed Items to Improve Long-Tail Performance](https://mlsanity.com/paper/2005.10545/) (2020)
- [Incorporating Bias-aware Margins into Contrastive Loss for Collaborative Filtering](https://mlsanity.com/paper/2210.11054/) (2022)
- [Related Pins at Pinterest: The Evolution of a Real-World Recommender System](https://mlsanity.com/paper/1702.07969/) (2017)
- [IntTower: the Next Generation of Two-Tower Model for Pre-Ranking System](https://mlsanity.com/paper/2210.09890/) (2022)
- [OmniSearchSage: Multi-Task Multi-Entity Embeddings for Pinterest Search](https://mlsanity.com/paper/2404.16260/) (2024)
- [A Multimodal Single-Branch Embedding Network for Recommendation in Cold-Start and Missing Modality Scenarios](https://mlsanity.com/paper/2409.17864/) (2024)
- [Connecting User and Item Perspectives in Popularity Debiasing for Collaborative Recommendation](https://mlsanity.com/paper/2006.04275/) (2020)
- [Learning Transferrable Parameters for Long-tailed Sequential User Behavior Modeling](https://mlsanity.com/paper/2010.11401/) (2020)
- [Cold & Warm Net: Addressing Cold-Start Users in Recommender Systems](https://mlsanity.com/paper/2309.15646/) (2023)
- [Latent User Intent Modeling for Sequential Recommenders](https://mlsanity.com/paper/2211.09832/) (2022)
- [Scene-wise Adaptive Network for Dynamic Cold-start Scenes Optimization in CTR Prediction](https://mlsanity.com/paper/2408.07278/) (2024)
- [Bootstrapping Contrastive Learning Enhanced Music Cold-Start Matching](https://mlsanity.com/paper/2308.02844/) (2023)

## Built on by (most cited first, 2 of 2)

- [PinEqualizer: Full Funnel Content Exploration and Debiasing System at Pinterest](https://mlsanity.com/paper/2607.22518/) (2026)
- [FORTE: A Production Framework for Feature Evaluation and Deprecation in Industrial Recommender Systems](https://mlsanity.com/paper/2610.13806/) (2026)

## Abstract (arXiv)

Pinterest is a leading visual discovery platform where recommender systems (RecSys) are key to delivering relevant, engaging, and fresh content to our users. In this paper, we study the problem of improving RecSys model predictions for cold-start (CS) items, which appear infrequently in the training data. Although this problem is well-studied in academia, few studies have addressed its root causes effectively at the scale of a platform like Pinterest. By investigating live traffic data, we identified several challenges of the CS problem and developed a corresponding solution for each: First, industrial-scale RecSys models must operate under tight computational constraints. Since CS items are a minority, any related improvements must be highly cost-efficient. To address this, our solutions were designed to be lightweight, collectively increasing the total parameters by only 5%. Second, CS items are represented only by non-historical (e.g., content or attribute) features, which models often treat as less important. To elevate their significance, we introduce a residual connection for the non-historical features. Third, CS items tend to receive lower prediction scores compared to non-CS items, reducing their likelihood of being surfaced. We mitigate this by incorporating a score regularization term into the model. Fourth, the labels associated with CS items are sparse, making it difficult for the model to learn from them. We apply the manifold mixup technique to address this data sparsity. Implemented together, our methods increased fresh content engagement at Pinterest by 10% without negatively impacting overall engagement and cost, and have been deployed to serve over 570 million users on Pinterest.