arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.08466cs.AIcs.CLcs.LG
收藏

用文字解释数据集:带自然语言参数的统计模型

Explaining Datasets in Words: Statistical Models with Natural Language Parameters

UC Berkeley(加州大学伯克利分校)

Ruiqi Zhong, Heng Wang, Dan Klein, Jacob Steinhardt

AI总结:

针对传统统计模型参数高维难解释的问题,提出由自然语言谓词参数化的统计模型族及模型无关学习算法,可应用于多领域多任务,解释复杂概念的能力优于经典方法。

摘要:

为理解海量数据,我们通常会拟合简化模型再解释其参数,例如对文本嵌入进行聚类,再解释每个簇的均值参数。但这些参数往往维度高、难以解读。为让模型参数具备直接可解释性,我们提出一类由自然语言谓词参数化的统计模型,涵盖聚类、时间序列和分类模型。例如,一个关于新冠的文本簇可由谓词“discusses COVID(讨论新冠)”参数化。为有效学习这些统计模型,我们开发了一种模型无关算法,通过梯度下降优化谓词参数的连续松弛形式,并通过提示语言模型(LM)对其进行离散化。最后,我们将该框架应用于多种问题:对用户聊天对话分类、刻画其随时间的演变规律、找出某一语言模型表现优于另一模型的类别、基于子领域对数学题聚类,以及解释难忘图像中的视觉特征。该框架通用性极强,可应用于文本和视觉领域,能轻松引导其聚焦特定属性(如子领域),还能解释经典方法(如n-gram分析)难以生成的复杂概念。

英文摘要:

To make sense of massive data, we often fit simplified models and then interpret the parameters; for example, we cluster the text embeddings and then interpret the mean parameters of each cluster. However, these parameters are often high-dimensional and hard to interpret. To make model parameters directly interpretable, we introduce a family of statistical models -- including clustering, time series, and classification models -- parameterized by natural language predicates. For example, a cluster of text about COVID could be parameterized by the predicate "discusses COVID". To learn these statistical models effectively, we develop a model-agnostic algorithm that optimizes continuous relaxations of predicate parameters with gradient descent and discretizes them by prompting language models (LMs). Finally, we apply our framework to a wide range of problems: taxonomizing user chat dialogues, characterizing how they evolve across time, finding categories where one language model is better than the other, clustering math problems based on subareas, and explaining visual features in memorable images. Our framework is highly versatile, applicable to both textual and visual domains, can be easily steered to focus on specific properties (e.g. subareas), and explains sophisticated concepts that classical methods (e.g. n-gram analysis) struggle to produce.

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。

↑