Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916938618241024 |
|---|---|
| author | Penaloza, Emiliano Zhang, Tianyue H. Charlin, Laurent Zarlenga, Mateo Espinosa |
| author_facet | Penaloza, Emiliano Zhang, Tianyue H. Charlin, Laurent Zarlenga, Mateo Espinosa |
| contents | Concept Bottleneck Models (CBMs) propose to enhance the trustworthiness of AI systems by constraining their decisions on a set of human-understandable concepts. However, CBMs typically assume that datasets contain accurate concept labels-an assumption often violated in practice, which we show can significantly degrade performance (by 25% in some cases). To address this, we introduce the Concept Preference Optimization (CPO) objective, a new loss function based on Direct Preference Optimization, which effectively mitigates the negative impact of concept mislabeling on CBM performance. We provide an analysis of key properties of the CPO objective, showing it directly optimizes for the concept's posterior distribution, and contrast it against Binary Cross Entropy (BCE), demonstrating that CPO is inherently less sensitive to concept noise. We empirically confirm our analysis by finding that CPO consistently outperforms BCE on three real-world datasets, both with and without added label noise. We make our code available on Github. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_18026 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization Penaloza, Emiliano Zhang, Tianyue H. Charlin, Laurent Zarlenga, Mateo Espinosa Machine Learning Artificial Intelligence Concept Bottleneck Models (CBMs) propose to enhance the trustworthiness of AI systems by constraining their decisions on a set of human-understandable concepts. However, CBMs typically assume that datasets contain accurate concept labels-an assumption often violated in practice, which we show can significantly degrade performance (by 25% in some cases). To address this, we introduce the Concept Preference Optimization (CPO) objective, a new loss function based on Direct Preference Optimization, which effectively mitigates the negative impact of concept mislabeling on CBM performance. We provide an analysis of key properties of the CPO objective, showing it directly optimizes for the concept's posterior distribution, and contrast it against Binary Cross Entropy (BCE), demonstrating that CPO is inherently less sensitive to concept noise. We empirically confirm our analysis by finding that CPO consistently outperforms BCE on three real-world datasets, both with and without added label noise. We make our code available on Github. |
| title | Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2504.18026 |