Causally Reliable Concept Bottleneck Models

2025-03-06

AI Summary

Current concept-based deep learning models lack explicit modeling of genuine causal mechanisms, limiting their applicability in causal reasoning, out-of-distribution generalization, and fairness-aware control. To address this, we propose a causally reliable and interpretable model: the first to embed a Structural Causal Model (SCM) into a concept bottleneck architecture, coupled with an end-to-end framework that jointly learns a causal concept graph from observational data and unstructured scientific literature. Our method integrates causal discovery, concept bottleneck networks, knowledge distillation, graph neural networks, text embeddings, and causal-constraint regularization. Experiments across multiple benchmarks demonstrate that our model matches black-box models in predictive accuracy while substantially improving intervention responsiveness (+27%), causal faithfulness (+34%), and human concept alignment. This work advances interpretable AI by grounding explanations in causally credible structures.

Technology Category

• Machine Learning: Causal Learning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Reasoning under Uncertainty: Causality

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Concept-based models are an emerging paradigm in deep learning that constrains the inference process to operate through human-interpretable concepts, facilitating explainability and human interaction. However, these architectures, on par with popular opaque neural models, fail to account for the true causal mechanisms underlying the target phenomena represented in the data. This hampers their ability to support causal reasoning tasks, limits out-of-distribution generalization, and hinders the implementation of fairness constraints. To overcome these issues, we propose emph{Causally reliable Concept Bottleneck Models} (C$^2$BMs), a class of concept-based architectures that enforce reasoning through a bottleneck of concepts structured according to a model of the real-world causal mechanisms. We also introduce a pipeline to automatically learn this structure from observational data and emph{unstructured} background knowledge (e.g., scientific literature). Experimental evidence suggest that C$^2$BM are more interpretable, causally reliable, and improve responsiveness to interventions w.r.t. standard opaque and concept-based models, while maintaining their accuracy.

Problem

• Enhance interpretability and causal reliability in concept-based models.
• Improve out-of-distribution generalization and fairness in AI models.
• Automate learning of causal structures from observational data and literature.

Innovation

• Enforces reasoning through causal concept bottlenecks
• Automatically learns causal structure from data
• Improves interpretability and causal reliability