Recent advancements in interpretable AI are focusing on enhancing the transparency of complex models, particularly in vision-language tasks and neural networks. New frameworks are emerging that allow for the extraction of human-interpretable concepts, enabling fine-grained explanations and spatial grounding in visual data. For instance, recent work has introduced models that not only map inputs to understandable concepts but also respect inter-concept relationships, reducing the need for extensive annotations. This shift towards hierarchical and causal structures facilitates better understanding and debugging of AI systems, addressing significant challenges in model interpretability. Additionally, the introduction of novel learning paradigms, such as Teleodynamic Learning, emphasizes the dynamic nature of intelligence, allowing models to adapt and self-organize while producing interpretable outputs. These developments have the potential to solve commercial problems in sectors like healthcare and autonomous systems, where understanding model decisions is crucial for trust and accountability.
Language-aligned vision foundation models perform strongly across diverse downstream tasks. Yet, their learned representations remain opaque, making interpreting their decision-making hard. Recent wor...
Deep neural networks (DNNs) achieve remarkable predictive performance but remain difficult to interpret, largely due to overparameterization that obscures the minimal structure required for interpreta...
We introduce Teleodynamic Learning, a new paradigm for machine learning in which learning is not the minimization of a fixed objective, but the emergence and stabilization of functional organization u...
Symbolic discovery of governing equations is a long-standing goal in scientific machine learning, yet a fundamental trade-off persists between interpretability and scalable learning. Classical symboli...
Concept Bottleneck Models (CBMs) enhance the interpretability of end-to-end neural networks by introducing a layer of concepts and predicting the class label from the concept predictions. A key proper...
Modern deep neural networks remain challenging to interpret due to the opacity of their latent representations, impeding model understanding, debugging, and debiasing. Concept Embedding Models (CEMs) ...
Learning PDE dynamics for fluids increasingly relies on neural operators and Transformer-based models, yet these approaches often lack interpretability and struggle with localized, high-frequency stru...