Neural-Symbolic Integration for Interactive Learning and Conceptual Grounding
This addresses the need for interpretability and user interaction in AI systems, particularly for domain experts, but it is incremental as it builds on existing neural-symbolic frameworks.
The paper tackles the problem of explaining and refining neural models through neural-symbolic integration, enabling interactive learning and conceptual grounding by allowing users to query models with symbolic logic and revise them based on logic-based constraints.
We propose neural-symbolic integration for abstract concept explanation and interactive learning. Neural-symbolic integration and explanation allow users and domain-experts to learn about the data-driven decision making process of large neural models. The models are queried using a symbolic logic language. Interaction with the user then confirms or rejects a revision of the neural model using logic-based constraints that can be distilled into the model architecture. The approach is illustrated using the Logic Tensor Network framework alongside Concept Activation Vectors and applied to a Convolutional Neural Network.