paper-with-me

홈 › Papers

Tackling Shortcut Learning in Deep Neural Networks: An Iterative Approach with Interpretable Models

2023-02-20 · Shantanu Ghosh, Ke Yu, Forough Arabshahi, Kayhan Batmanghelich

We use concept-based interpretable models to mitigate shortcut learning. Existing methods lack interpretability. Beginning with a Blackbox, we iteratively carve out a mixture of interpretable experts (MoIE) and a residual network. Each expert explains a subset of data using First Order Logic (FOL). While explaining a sample, the FOL from biased BB-derived MoIE detects the shortcut effectively. Finetuning the BB with Metadata Normalization (MDN) eliminates the shortcut. The FOLs from the finetuned-BB-derived MoIE verify the elimination of the shortcut. Our experiments show that MoIE does not hurt the accuracy of the original BB and eliminates shortcuts effectively.

📄 PDF Abstract BibTeX arXiv:2302.10289

Code (1)

batmanlab/ICML-2023-Route-interpret-repeat 공식 구현 pytorch

Methods 이 논문이 사용한 방법론

HOC 설명 없음

Similar Papers 제목 키워드 기반

Patch Shortcuts: Interpretable Proxy Models Efficiently Find Black-Box Vulnerabilities

2021-04-22 · Julia Rosenzweig, Joachim Sicking, Sebastian Houben, Michael Mock 외

An important pillar for safe machine learning (ML) is the systematic mitigation of weaknesses in neural networks to afford their deployment in critical applications. An ubiquitous class of safety risks are learned shortc…

Autonomous Driving

Dividing and Conquering a BlackBox to a Mixture of Interpretable Models: Route, Interpret, Repeat

2023-07-07 · Shantanu Ghosh, Ke Yu, Forough Arabshahi, Kayhan Batmanghelich

ML model design either starts with an interpretable model or a Blackbox and explains it post hoc. Blackbox models are flexible but difficult to explain, while interpretable models are inherently explainable. Yet, interpr…

ShortcutLens: A Visual Analytics Approach for Exploring Shortcuts in Natural Language Understanding Dataset

2022-08-17 · Zhihua Jin, Xingbo Wang, Furui Cheng, Chunhui Sun 외

Benchmark datasets play an important role in evaluating Natural Language Understanding (NLU) models. However, shortcuts -- unwanted biases in the benchmark datasets -- can damage the effectiveness of benchmark datasets i…

Natural Language Understanding

Generative Poisoning Using Random Discriminators

2022-11-02 · Dirren van Vlijmen, Alex Kolmus, Zhuoran Liu, Zhengyu Zhao 외

We introduce ShortcutGen, a new data poisoning attack that generates sample-dependent, error-minimizing perturbations by learning a generator. The key novelty of ShortcutGen is the use of a randomly-initialized discrimin…

Data Poisoning

The Devil Is in the Details: Tackling Unimodal Spurious Correlations for Generalizable Multimodal Reward Models

2025-03-05 · Zichao Li, Xueru Wen, Jie Lou, Yuqiu Ji 외

Multimodal Reward Models (MM-RMs) are crucial for aligning Large Language Models (LLMs) with human preferences, particularly as LLMs increasingly interact with multimodal data. However, we find that MM-RMs trained on exi…