标题： 基于机器学习的无信号假设的新物理多重检验 显示英文标题

标题： Multiple testing for signal-agnostic searches of new physics with machine learning

摘要： 在这项工作中，我们解决了如何通过利用多种检验策略来增强信号无关搜索的问题。 具体而言，我们考虑依赖于机器学习的假设检验，其中模型选择可能会导致对特定新物理信号族的偏差。 我们表明，结合由不同超参数选择表征的检验是有益的，并且通常可以实现与现有最佳检验相当的性能，同时对各种类型的异常提供更均匀的响应。 专注于新物理学习机，这是一种执行信号无关似然比检验的方法，我们探讨了多种多重检验的方法，例如结合p值和聚合检验统计量。 显示英文摘要

摘要： In this work, we address the question of how to enhance signal-agnostic searches by leveraging multiple testing strategies. Specifically, we consider hypothesis tests relying on machine learning, where model selection can introduce a bias towards specific families of new physics signals. We show that it is beneficial to combine different tests, characterised by distinct choices of hyperparameters, and that performances comparable to the best available test are generally achieved while providing a more uniform response to various types of anomalies. Focusing on the New Physics Learning Machine, a methodology to perform a signal-agnostic likelihood-ratio test, we explore a number of approaches to multiple testing, such as combining p-values and aggregating test statistics.