Generative Modeling with Multi-Instance Reward Learning for E-commerce Creative Optimization

2025-08-13

AI Summary

E-commerce advertising faces combinatorial explosion in creative elements (e.g., titles, images, highlights), rendering optimization intractable; existing methods typically evaluate components in isolation and struggle to efficiently search the exponential space of combinations. This paper proposes GenCO, a generative compositional optimization framework that integrates generative modeling with multi-instance reinforcement learning. Specifically, it employs a generative model to jointly sample high-quality, diverse creative combinations, while leveraging multi-instance learning to decompose组合-level click/conversion feedback into element-wise rewards—enabling fine-grained reward attribution and interpretable optimization. Deployed at scale on major e-commerce platforms, GenCO significantly improves CTR, CVR, and advertising revenue. Additionally, we release the first industrial-scale e-commerce creative composition dataset to foster community research.

Technology Category

• Search and Optimization: Combinatorial Optimization
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Search and Retrieval-Augmented AI: Ad search and search for Web retail

Abstract

In e-commerce advertising, selecting the most compelling combination of creative elements -- such as titles, images, and highlights -- is critical for capturing user attention and driving conversions. However, existing methods often evaluate creative components individually, failing to navigate the exponentially large search space of possible combinations. To address this challenge, we propose a novel framework named GenCO that integrates generative modeling with multi-instance reward learning. Our unified two-stage architecture first employs a generative model to efficiently produce a diverse set of creative combinations. This generative process is optimized with reinforcement learning, enabling the model to effectively explore and refine its selections. Next, to overcome the challenge of sparse user feedback, a multi-instance learning model attributes combination-level rewards, such as clicks, to the individual creative elements. This allows the reward model to provide a more accurate feedback signal, which in turn guides the generative model toward creating more effective combinations. Deployed on a leading e-commerce platform, our approach has significantly increased advertising revenue, demonstrating its practical value. Additionally, we are releasing a large-scale industrial dataset to facilitate further research in this important domain.

Problem

• Optimizing e-commerce creative combinations for user engagement
• Addressing sparse feedback in evaluating creative elements
• Navigating large search space of creative combinations

Innovation

• Generative model produces diverse creative combinations
• Reinforcement learning optimizes creative selection process
• Multi-instance learning attributes rewards to elements