Abstract

In the last few years, Deep Learning (DL) has achieved impressive results in a variety of problems ranging from computer vision to natural language processing. Nonetheless, the excitement around the field may remain disappointed since there are still many open issues. To mitigate some of these problems, we consider the Learning from Constraints framework. In this setting learning is conceived as the problem of finding task functions while respecting the constraints representing the available knowledge. We provide an application in the Active Learning scenario where First-Order Logic knowledge is converted into constraints and their violation is checked as a guide for sample selection. Also, we propose to employ domain knowledge to defend from Adversarial Attacks since it provides a natural guide to detect adversarial examples. While some relationships are known properties of the considered environments, DNNs can also autonomously develop new relation patterns. Therefore, we also propose a novel Learning of Constraints formulation which aims at understanding which logic constraints are satisfied by the task functions. This allows explaining DNNs, otherwise commonly considered black-box classifiers. In a first case, we propose a pair of neural networks, where one learns the relationships among the outputs of the other one and provides First-Order Logic (FOL)-based descriptions. In a second case, we propose an end-to-end differentiable approach, extracting logic explanations from the same classifier. The method relies on an entropy-based layer which automatically identifies the most relevant concepts. It enables the distillation of concise logic explanations in several safety-critical domains, outperforming state-of-the-art white-box models.

https://cl-cam-ac-uk.zoom.us/j/99805544705?pwd=cXR6MTlaeXd6VmEreVdQSmFRblBtUT09