A New Creative Generation Pipeline for Click-Through Rate with Stable Diffusion Model

Abstract

In online advertising scenario, sellers often create multiple creatives to provide comprehensive demonstrations, making it essential to present the most appealing design to maximize the Click-Through Rate (CTR). However, sellers generally struggle to consider users’ preferences for creative design, leading to the relatively lower aesthetics and quantities compared to Artificial Intelligence (AI)-based approaches. Traditional AI-based approaches still face the same problem of not considering user information while having limited aesthetic knowledge from designers. In fact that fusing the user information, the generated creatives can be more attractive because different users may have different preferences. To optimize the results, the generated creatives in traditional methods are then ranked by another module named creative ranking model. The ranking model can predict the CTR score for each creative considering user features. However, the two above stages (generating creatives and ranking creatives) are regarded as two different tasks and are optimized separately. Specifically, generating creatives in the first stage without considering the target of improving CTR task may generate several creatives with poor quality, leading to dilute online impressions and directly making bad effectiveness on online results. In this paper, we proposed a new automated Creative Generation pipeline for Click-Through Rate (CG4CTR) 1 with the goal of improving CTR during the creative generation stage. In this pipeline, a new creative is automatically generated and selected by stable diffusion method with the LoRA model and two novel models: prompt model and reward model. Our contributions have four parts: 1) The inpainting mode in stable diffusion method is firstly applied to creative image generation task in online advertising scene. A self-cyclic generation pipeline is proposed to ensure the convergence of training. 2) Prompt model is designed to generate individualized creative images for different user groups, which can further improve the diversity and quality of the generated creatives. 3) Reward model comprehensively considers the multi-modal features of image and text to improve the effectiveness of creative ranking task, and it is also critical in self-cyclic generation pipeline. 4) The significant benefits obtained in online and offline experiments verify the significance of our proposed method.