Presenting a Paper is an Art: Self-Improvement Aesthetic Agents for Academic Presentations
Chengzhi Liu, Yuzhe Yang, Kaiwen Zhou, Zhen Zhang, Yue Fan, Yannan Xie, Peng Qi, Xin Eric Wang
2025-10-08
Summary
This paper introduces a new system called EvoPresent that aims to automatically create better presentations for academic research papers, making them more engaging and visually appealing.
What's the problem?
Currently, automated tools for creating presentations from research papers aren't very good. They often struggle to tell a compelling story, the designs aren't aesthetically pleasing, and they can't really improve themselves without human input. The core issue is that these systems can't accurately judge how good their presentations are, so they can't learn to make them better. It's hard to fix something if you can't evaluate it!
What's the solution?
The researchers developed EvoPresent, which uses a technique called reinforcement learning to create and refine presentations. A key part of EvoPresent is a model called PresAesth that's specifically designed to evaluate the visual quality of slides – things like color schemes, layout, and overall design. PresAesth can score presentations, identify flaws, and compare different designs, even when there isn't a lot of example data to learn from. They also created a large dataset of research papers and presentations to test their system thoroughly.
Why it matters?
This work is important because it addresses a real need in the academic community: effectively sharing research findings. Better automated presentations can help researchers reach a wider audience and have a greater impact. The findings show that good feedback is crucial for improvement, and that balancing visual design with strong content is a challenge for these automated systems. Ultimately, this research moves us closer to AI tools that can create truly effective and engaging presentations without constant human intervention.
Abstract
The promotion of academic papers has become an important means of enhancing research visibility. However, existing automated methods struggle limited storytelling, insufficient aesthetic quality, and constrained self-adjustment, making it difficult to achieve efficient and engaging dissemination. At the heart of those challenges is a simple principle: there is no way to improve it when you cannot evaluate it right. To address this, we introduce EvoPresent, a self-improvement agent framework that unifies coherent narratives, aesthetic-aware designs, and realistic presentation delivery via virtual characters. Central to EvoPresent is PresAesth, a multi-task reinforcement learning (RL) aesthetic model that provides reliable aesthetic scoring, defect adjustment, and comparative feedback, enabling iterative self-improvement even under limited aesthetic training data. To systematically evaluate the methods, we introduce EvoPresent Benchmark, a comprehensive benchmark comprising: Presentation Generation Quality, built on 650 top-tier AI conference papers with multimodal resources (slides, videos and scripts) to assess both content and design; and Aesthetic Awareness, consisting of 2,000 slide pairs with varying aesthetic levels, supporting joint training and evaluation on scoring, defect adjustment, and comparison. Our findings highlight that (i) High-quality feedback is essential for agent self-improvement, while initial capability alone does not guarantee effective self-correction. (ii) Automated generation pipelines exhibit a trade-off between visual design and content construction. (iii) Multi-task RL training shows stronger generalization in aesthetic awareness tasks.