Augmenting deep neural networks with symbolic knowledge: Towards trustworthy and interpretable AI for education

TL;DR

The paper addresses the need for trustworthy and interpretable AI in education by augmenting deep neural networks with symbolic knowledge through a neural-symbolic NSAI framework. NSAI injects propositional educational knowledge into network architecture using an adapted KBANN approach, trains with data, augments with synthetic samples, and extracts rule-based explanations to justify predictions. On the AutoThinking CT game dataset, NSAI demonstrates superior generalizability and interpretability over data-only and data-augmented baselines, while mitigating spurious correlations via explicit knowledge constraints. The results suggest that neural-symbolic integration can yield educational AI that is more robust, fair, and explainable, with potential for refinement of domain knowledge and practical deployment in classrooms.

Abstract

Artificial neural networks (ANNs) have shown to be amongst the most important artificial intelligence (AI) techniques in educational applications, providing adaptive educational services. However, their educational potential is limited in practice due to three major challenges: i) difficulty in incorporating symbolic educational knowledge (e.g., causal relationships, and practitioners' knowledge) in their development, ii) learning and reflecting biases, and iii) lack of interpretability. Given the high-risk nature of education, the integration of educational knowledge into ANNs becomes crucial for developing AI applications that adhere to essential educational restrictions, and provide interpretability over the predictions. This research argues that the neural-symbolic family of AI has the potential to address the named challenges. To this end, it adapts a neural-symbolic AI framework and accordingly develops an approach called NSAI, that injects and extracts educational knowledge into and from deep neural networks, for modelling learners computational thinking. Our findings reveal that the NSAI approach has better generalizability compared to deep neural networks trained merely on training data, as well as training data augmented by SMOTE and autoencoder methods. More importantly, unlike the other models, the NSAI approach prioritises robust representations that capture causal relationships between input features and output labels, ensuring safety in learning to avoid spurious correlations and control biases in training data. Furthermore, the NSAI approach enables the extraction of rules from the learned network, facilitating interpretation and reasoning about the path to predictions, as well as refining the initial educational knowledge. These findings imply that neural-symbolic AI can overcome the limitations of ANNs in education, enabling trustworthy and interpretable applications.

Figures (7)

• Figure 1: (a) a solution developed by a player, and (b) feedback and a hint generated for the solution (taken from Hooshyar et al., hooshyar2021gaming).
• Figure 2: The overall architecture of the NSAI approach.
• Figure 3: (a) initializing the network using the domain knowledge, and (b) the adjusted network after training.
• Figure 4: Deviation chart showing the distribution of: a) the original training, b) training augmented with SMOTE, c) training augmented with Autoencoder, and d) test data.
• Figure 5: Global LIME explanations, a) Deep NN, b) Deep NN-SMOTE, and c) Deep NN-Autoencoder.