CTBENCH：一个用于认证训练的库与基准
CTBENCH: A Library and Benchmark for Certified Training
AI总结：
本文提出CTBENCH，一个统一的认证训练库与基准，通过公平设置和系统调优超参数评估算法，发现近期算法优势在公平比较下显著下降，并揭示了认证模型损失表面、激活稀疏性等新特性。
摘要：
训练可认证鲁棒的神经网络是一项重要但具有挑战性的任务。尽管已经提出了许多用于（确定性）认证训练的算法，但它们通常在不同的训练计划、认证方法和系统性的欠调超参数下进行评估，这使得比较它们的性能变得困难。为了解决这一挑战，我们引入了CTBENCH，这是一个统一的库和高质量的认证训练基准，它在公平的设置和系统调优的超参数下评估所有算法。我们表明，（1）CTBENCH中几乎所有算法在算法改进幅度上都超过了文献中报告的相应性能，从而建立了新的最先进水平，以及（2）当我们用公平的训练计划、公平的认证方法和良好调优的超参数来增强过时的基线时，最近算法所声称的优势显著下降。基于CTBENCH，我们对认证训练的现状提供了新的见解，包括（1）认证模型具有较少碎片化的损失表面，（2）认证模型共享许多错误，（3）认证模型具有更稀疏的激活，（4）巧妙地减少正则化对于认证训练至关重要，特别是对于大半径，以及（5）认证训练有潜力改善分布外泛化。我们相信CTBENCH将作为未来认证训练研究的基准和测试平台。
英文摘要：
Training certifiably robust neural networks is an important but challenging task. While many algorithms for (deterministic) certified training have been proposed, they are often evaluated on different training schedules, certification methods, and systematically under-tuned hyperparameters, making it difficult to compare their performance. To address this challenge, we introduce CTBench, a unified library and a high-quality benchmark for certified training that evaluates all algorithms under fair settings and systematically tuned hyperparameters. We show that (1) almost all algorithms in CTBench surpass the corresponding reported performance in literature in the magnitude of algorithmic improvements, thus establishing new state-of-the-art, and (2) the claimed advantage of recent algorithms drops significantly when we enhance the outdated baselines with a fair training schedule, a fair certification method and well-tuned hyperparameters. Based on CTBench, we provide new insights into the current state of certified training, including (1) certified models have less fragmented loss surface, (2) certified models share many mistakes, (3) certified models have more sparse activations, (4) reducing regularization cleverly is crucial for certified training especially for large radii and (5) certified training has the potential to improve out-of-distribution generalization. We are confident that CTBench will serve as a benchmark and testbed for future research in certified training.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
