arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.01408cs.LGcs.AI
收藏

反事实概念瓶颈模型

Counterfactual Concept Bottleneck Models

Università della Svizzera italiana(瑞士意大利语大学);IBM Research(IBM研究院);Scuola Normale Superiore(高等师范学院);KU Leuven(鲁汶大学)

Gabriele Dominici, Pietro Barbiero, Francesco Giannini, Martin Gjoreski, Giuseppe Marra, Marc Langheinrich

AI总结:

本文提出CF-CBMs,通过联合训练反事实生成器与概念瓶颈模型,同时实现高准确率分类、简洁概念解释和可解释反事实生成。

摘要:

当前的深度学习模型并非被设计为同时解决三个基本问题:预测类别标签以完成给定分类任务(“是什么?”)、模拟情境变化以评估其对类别预测的影响(“如何?”),以及设想情境应如何改变以产生不同的类别预测(“为什么不?”)。无法回答这些问题,是在部署可靠AI智能体、校准人类信任以及改善人机交互方面的一个关键缺口。为弥合这一缺口,我们提出了反事实概念瓶颈模型(CounterFactual Concept Bottleneck Models,CF-CBMs),这是一类旨在一次性高效处理上述查询、而无需运行事后搜索的模型。实验结果表明,CF-CBMs:取得了与黑盒模型和现有CBMs相当的分类准确率(“是什么?”),依赖更少的重要概念,从而产生更简单的解释(“如何?”),并生成可解释的、基于概念的反事实(“为什么不?”)。此外,我们表明,将反事实生成器与CBM联合训练会带来两项关键改进:(i)它改变了模型的决策过程,使模型依赖更少的重要概念(从而产生更简单的解释);(ii)它显著增强了概念干预对类别预测的因果效应,使模型对这些变化更敏感。

英文摘要:

Current deep learning models are not designed to simultaneously address three fundamental questions: predict class labels to solve a given classification task (the "What?"), simulate changes in the situation to evaluate how this impacts class predictions (the "How?"), and imagine how the scenario should change to result in different class predictions (the "Why not?"). The inability to answer these questions represents a crucial gap in deploying reliable AI agents, calibrating human trust, and improving human-machine interaction. To bridge this gap, we introduce CounterFactual Concept Bottleneck Models (CF-CBMs), a class of models designed to efficiently address the above queries all at once without the need to run post-hoc searches. Our experimental results demonstrate that CF-CBMs: achieve classification accuracy comparable to black-box models and existing CBMs ("What?"), rely on fewer important concepts leading to simpler explanations ("How?"), and produce interpretable, concept-based counterfactuals ("Why not?"). Additionally, we show that training the counterfactual generator jointly with the CBM leads to two key improvements: (i) it alters the model's decision-making process, making the model rely on fewer important concepts (leading to simpler explanations), and (ii) it significantly increases the causal effect of concept interventions on class predictions, making the model more responsive to these changes.

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。

↑