Multi-Valued Autoencoders and Classification of Large-Scale Multi-Class Problem

Abstract

Two-layered neural networks are well known as autoencoders (AEs) in order to reduce the dimensionality of data. AEs are successfully employed as pre-trained layers of neural networks for classification tasks. Most of the existing studies conceived real-valued AEs in real-valued neural networks. This study investigated complex- and quaternion-valued AEs for complex- and quaternion-valued neural networks. Inputs, weights, biases, and outputs in complex-valued AE (CAE) are complex variables, whereas those in quaternion-valued AE (QAE) are quaternions. In both methods, a split-type activation function is used in the hidden and output units. To deal with the images using the proposed methods, pairs of pixels are allotted to complex-valued inputs in the CAE and quartets of pixels are allotted to quaternion-valued inputs in the QAE. Proposed autoencoders are tested and performance compared with conventional AE for several tasks which are encoding/decoding, handwritten numeral recognition and large-scale multi-class classification. Proposed CAE and QAE revealed as good recognition methods for the tasks and outperformed conventional AE with significance performance in case of large-scale multi-class images recognition.

Authors and Affiliations

Ryusuke Hata, M. A. H. Akhand, Kazuyuki Murase

Keywords

Related Articles

Cross Language Information Retrieval Model for Discovering WSDL Documents Using Arabic Language Query

Web service discovery is the process of finding a suitable Web service for a given user’s query through analyzing the web service‘s WSDL content and finding the best match for the user’s query. The service query should b...

AL-S[sup]2[/sup]m: Soft road traffic Signs map for vehicular systems

In this paper, we describe AL-S[sup]2[/sup]m, a roadmap with traffic signs to be used in vehicular systems. AL-S[sup]2[/sup]m is part of a more general system of traffic signs (TSs) management, called AL-S[sup]2[/sup], w...

Comparison and Analysis of Different Software Cost Estimation Methods

Software cost estimation is the process of predicting the effort required to develop a software system. The basic input for the software cost estimation is coding size and set of cost drivers, the output is Effort in ter...

A Novel Multiple Session Payment System

A wireless smartphone can be designed to process a financial payment efficiently. A user can just swipe his/her credit/debit card over the counter and all the processing needed shall be done seamlessly. A smartphone is a...

Smart Surveillance System using Background Subtraction Technique in IoT Application

This paper presents a development of a security system based on Internet-of-Things (IoT) technology, where an IoT device, Raspberry Pi has been used. In the developed surveillance system, a camera works as a sensor to de...

Download PDF file
  • EP ID EP240416
  • DOI 10.14569/IJACSA.2017.081103
  • Views 87
  • Downloads 0

How To Cite

Ryusuke Hata, M. A. H. Akhand, Kazuyuki Murase (2017). Multi-Valued Autoencoders and Classification of Large-Scale Multi-Class Problem. International Journal of Advanced Computer Science & Applications, 8(11), 19-26. https://europub.co.uk/articles/-A-240416