Concept Based Approach

Concept-based approaches aim to improve the interpretability and performance of artificial neural networks by representing model decisions in terms of human-understandable concepts. Current research focuses on developing methods for automatically extracting and evaluating these concepts, often within a federated learning framework to handle non-IID data, and using them to refine model architectures and improve generalization. This work is significant for advancing explainable AI (XAI), enabling better understanding of model behavior, identifying biases, and ultimately leading to more trustworthy and reliable AI systems across various applications.