[论文] Procedural Adversarial Learning

论文概要

研究领域: CV 作者: Jack Nugent, Jia Deng 发布时间: 2026-11-04 arXiv: 2611.06890

中文摘要

对许多计算机视觉任务，神经网络可被训练到产生较低平均误差，但仍在看似简单的场景上失败。减少这些失败需先发现它们，但它们可能不在固定数据集中。我们研究程序化对抗学习（PAL）——一个极小极大训练和对抗评估框架，在程序化生成器的输入空间中搜索困难实例。程序化生成器是可解释程序，将低维控制参数向量（布局、光照、姿态等）映射到渲染图像及其标签。我们提出 Simple-PAT 算法，在程序化生成器参数空间中使用零阶优化按需生成对抗训练样本。我们还创建 PAL-Dojo——面向表面法线估计、图像分割和人体姿态估计的高效程序化生成器集合。对抗 5000 万次聚合查询的跨模型黑盒对手时，Simple-PAT 将最坏情况误差相比标准训练降低几何平均 3.15 倍，相比最强鲁棒训练基线降低 2.63 倍。代码：https://github.com/princeton-vl/procedural-adversarial-learning。

原文摘要

For many computer vision tasks, neural networks can be trained to produce a low average error, yet they still fail on seemingly simple scenes. Reducing these failures requires first finding them, but they may not be in a fixed dataset. We study procedural adversarial learning (PAL), a framework of minimax training and adversarial evaluation which searches the input space of a procedural generator for difficult instances. The procedural generator is an interpretable program mapping a low-dimensional vector of control parameters (e.g. layout, lighting, pose) to a rendered image and its label. We introduce the Simple-PAT algorithm, which uses zeroth-order optimization in the procedural generator's parameter space to propose adversarial training samples that are generated on demand. We also cr...

#论文 #arXiv #CV #小凯