logic-based machine learning aims to learn general, interpretable knowledge in a data-efficient manner. However, labelled data must be specified in a structured logical form. To address this limitation, we propose a n...
详细信息
logic-based machine learning aims to learn general, interpretable knowledge in a data-efficient manner. However, labelled data must be specified in a structured logical form. To address this limitation, we propose a neural-symbolic learning framework, called Feed-Forward Neural-Symbolic Learner (FFNSL), that integrates a logic-based machine learning system capable of learning from noisy examples, with neural networks, in order to learn interpretable knowledge from labelled unstructured data. We demonstrate the generality of FFNSL on four neural-symbolic classification problems, where different pre-trained neural network models and logic-based machine learning systems are integrated to learn interpretable knowledge from sequences of images. We evaluate the robustness of our framework by using images subject to distributional shifts, for which the pre-trained neural networks may predict incorrectly and with high confidence. We analyse the impact that these shifts have on the accuracy of the learned knowledge and run-time performance, comparing FFNSL to tree-based and pure neural approaches. Our experimental results show that FFNSL outperforms the baselines by learning more accurate and interpretable knowledge with fewer examples.
This paper presents an interactive system for exploring and editing logic-based machine learning models specialised for the relational reasoning problem domain. Prior work has highlighted the value of visual interface...
详细信息
ISBN:
(纸本)9798400701078
This paper presents an interactive system for exploring and editing logic-based machine learning models specialised for the relational reasoning problem domain. Prior work has highlighted the value of visual interfaces for enabling effective user interaction during model training. However, these existing systems require two-dimensional tabular data and are not well-suited to relational machinelearning tasks. logic-based methods, such as those developed in the field of Inductive logic Programming, can address this;they retain relational information by operating directly on raw relational data while remaining inherently interpretable and editable to allow for human intervention. However, such systems require logical expertise to operate effectively and do not enable visual exploration. We aim to address this;taking design inspiration from equivalent interfaces for propositional learning, we present a visual interface that enhances the usability of inductive logic programming systems for domain experts without a background in computational logic.
暂无评论