A Universal Logic Operator for Interpretable Deep Convolution Networks
This addresses the need for interpretability in deep learning, but appears incremental as it builds on existing logical operator frameworks.
The paper tackles the problem of learning a universal logical operator for interpretable deep convolutional networks without manual prescription, resulting in a novel logical interpretation for these networks.
Explaining neural network computation in terms of probabilistic/fuzzy logical operations has attracted much attention due to its simplicity and high interpretability. Different choices of logical operators such as AND, OR and XOR give rise to another dimension for network optimization, and in this paper, we study the open problem of learning a universal logical operator without prescribing to any logical operations manually. Insightful observations along this exploration furnish deep convolution networks with a novel logical interpretation.