Logic-Based Neural Network for Pattern Correction

Abstract

As the 21st century continues, Deep Learning (DL) has become an important part in the digital world. The use of Artificial Neural Networks (ANNs), the core of DL, has led to the development of many novel applications, like speech recognition, character recognition, autonomous cars, etc. In this paper, we present a logic-based neural network processor for pattern correction, which can detect anomalies in a certain character, and can regenerate the character after removing the anomalies. The processor constitutes of sub-networks, namely the Generator network, the Inverter network, the Locator network, the Identifier network, and the Replacer network. A test for the proposed processor was done on the dataset of numeric characters.