A parameterized activation function for learning fuzzy logic operations in deep neural networks

TL;DR

This paper introduces a novel parameterized activation function for deep neural networks that enables learning fuzzy logic operations, providing interpretability and versatility in modeling logical relationships.

Contribution

The paper proposes a new differentiable activation function that can learn multiple fuzzy logic operations, with a theoretical foundation and successful application to classification tasks.

Findings

Effective learning of fuzzy logic expressions in neural networks

Provides interpretability of logical relationships in learned models

Achieves competitive results on UCI classification datasets

Abstract

We present a deep learning architecture for learning fuzzy logic expressions. Our model uses an innovative, parameterized, differentiable activation function that can learn a number of logical operations by gradient descent. This activation function allows a neural network to determine the relationships between its input variables and provides insight into the logical significance of learned network parameters. We provide a theoretical basis for this parameterization and demonstrate its effectiveness and utility by successfully applying our model to five classification problems from the UCI Machine Learning Repository.