arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.02876cs.LGcs.AIstat.ML
收藏

用于可解释预测的变分信息追踪

Variational Information Pursuit for Interpretable Predictions

Johns Hopkins University(约翰斯·霍普金斯大学)

Aditya Chattopadhyay, Kwan Ho Ryan Chan, Benjamin D. Haeffele, Donald Geman, René Vidal

AI总结:

本文提出变分信息追踪(V-IP),通过最小化期望交叉熵直接学习最优查询选择策略和分类器,无需生成模型,在视觉、NLP和医学诊断任务上实现了更快推理、更短查询链和更优性能。

摘要:

机器学习社区对开发“设计上可解释”的预测算法越来越感兴趣。为此,近期工作提出通过顺序地提出关于数据的可解释查询来做出可解释决策,直到基于所获得的答案(历史记录)能够以高置信度做出预测。为了促进较短的查询-答案链,使用了一种称为信息追踪(IP)的贪心过程,该过程按信息增益顺序自适应地选择查询。生成模型被用来学习查询-答案和标签的分布,进而用于估计最具信息量的查询。然而,对于复杂任务,使用完整的数据生成模型进行学习和推理通常是不可行的。在这项工作中,我们提出了变分信息追踪(V-IP),这是IP的一种变分刻画,它绕过了学习生成模型的需要。V-IP基于寻找一个查询选择策略和一个分类器,以最小化真实标签与预测标签之间的期望交叉熵。我们随后证明IP策略是该问题的最优解。因此,我们可以使用我们的最优策略直接在任何给定历史记录下选择最具信息量的查询,而无需学习生成模型。然后,我们通过使用深度网络定义策略和分类器的有限维参数化,并利用我们的目标函数对它们进行端到端训练,开发了一个实用算法。在经验上,V-IP在不同的视觉和自然语言处理任务上比IP快10-100倍,且性能具有竞争力。此外,与通常用于顺序决策问题的强化学习相比,V-IP找到了更短的查询链。最后,我们展示了V-IP在诸如医学诊断等具有挑战性的任务上的实用性,其性能远优于生成建模方法。

英文摘要:

There is a growing interest in the machine learning community in developing predictive algorithms that are "interpretable by design". Towards this end, recent work proposes to make interpretable decisions by sequentially asking interpretable queries about data until a prediction can be made with high confidence based on the answers obtained (the history). To promote short query-answer chains, a greedy procedure called Information Pursuit (IP) is used, which adaptively chooses queries in order of information gain. Generative models are employed to learn the distribution of query-answers and labels, which is in turn used to estimate the most informative query. However, learning and inference with a full generative model of the data is often intractable for complex tasks. In this work, we propose Variational Information Pursuit (V-IP), a variational characterization of IP which bypasses the need for learning generative models. V-IP is based on finding a query selection strategy and a classifier that minimizes the expected cross-entropy between true and predicted labels. We then demonstrate that the IP strategy is the optimal solution to this problem. Therefore, instead of learning generative models, we can use our optimal strategy to directly pick the most informative query given any history. We then develop a practical algorithm by defining a finite-dimensional parameterization of our strategy and classifier using deep networks and train them end-to-end using our objective. Empirically, V-IP is 10-100x faster than IP on different Vision and NLP tasks with competitive performance. Moreover, V-IP finds much shorter query chains when compared to reinforcement learning which is typically used in sequential-decision-making problems. Finally, we demonstrate the utility of V-IP on challenging tasks like medical diagnosis where the performance is far superior to the generative modelling approach.

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。

↑