Adoption of the Activation Function Fusion Approach to Identify Human Activity Recognition in a Semi-Supervised Neural Network

Research output: Contribution to journalArticle

Abstract

INTRODUCTION: Neural networks are a popular type of algorithm for human activity monitoring which can build intelligent systems from labelled data in an automated fashion. Obtaining accurately labelled data is costly; it requires time and effort, which can be cumbersome because it interrupts the user activity stream. In conjunction with the ubiquitous presence of embedded technology, neural networks present new research opportunities for human activity monitoring in smart home environments. OBJECTIVES: We propose a human activity classification method that requires a limited amount of labelled data, which consists of a concatenation method for classifying human activities built upon the fusion of neural network activation functions. METHODS: Our methodology builds a neural network model that receives the sensor data through the input layer to then distribute it among the different vertical hidden layers, which implement different activation functions simultaneously. Next a hidden layer combines activation functions by utilising a concatenation method. Finally, the neural network provides classes to the unlabelled sensing data. We conducted an evaluation utilising an open-access dataset. We compared the activity recognition accuracy of our approach utilising 25%, 50%, and 75% of labelled data against a conventional shallow neural network trained with the 100% of labelled data available. RESULTS: Results show an improvement in the accuracy of the activity classification regardless of the portion of labelled data available. It was observed that the highest achieved accuracy when using 25% of activation function fusion data outperformed results compared to when using 100% of labelled data in a conventional shallow network (i.e., increase in accuracy of 2.7%, 3.7%, 4.8%, and 0.9% across the activity recognition of four subjects). CONCLUSION: The approach proposed showed an improvement in the accuracy of classifying human activity when a limited amount of labelled data is available.
LanguageEnglish
Article numbere3
Number of pages15
JournalEAI Endorsed Transactions on Pervasive Health and Technology
Volume5
Issue number17
DOIs
Publication statusPublished - 19 Nov 2019

Fingerprint

Chemical activation
Neural networks
Monitoring
Data fusion
Intelligent systems
Sensors

Keywords

  • Activation function
  • Activity recognition
  • Data fusion
  • Neural networks

Cite this

@article{ef17cc3b491a41cfa728728cb3fe3c50,
title = "Adoption of the Activation Function Fusion Approach to Identify Human Activity Recognition in a Semi-Supervised Neural Network",
abstract = "INTRODUCTION: Neural networks are a popular type of algorithm for human activity monitoring which can build intelligent systems from labelled data in an automated fashion. Obtaining accurately labelled data is costly; it requires time and effort, which can be cumbersome because it interrupts the user activity stream. In conjunction with the ubiquitous presence of embedded technology, neural networks present new research opportunities for human activity monitoring in smart home environments. OBJECTIVES: We propose a human activity classification method that requires a limited amount of labelled data, which consists of a concatenation method for classifying human activities built upon the fusion of neural network activation functions. METHODS: Our methodology builds a neural network model that receives the sensor data through the input layer to then distribute it among the different vertical hidden layers, which implement different activation functions simultaneously. Next a hidden layer combines activation functions by utilising a concatenation method. Finally, the neural network provides classes to the unlabelled sensing data. We conducted an evaluation utilising an open-access dataset. We compared the activity recognition accuracy of our approach utilising 25{\%}, 50{\%}, and 75{\%} of labelled data against a conventional shallow neural network trained with the 100{\%} of labelled data available. RESULTS: Results show an improvement in the accuracy of the activity classification regardless of the portion of labelled data available. It was observed that the highest achieved accuracy when using 25{\%} of activation function fusion data outperformed results compared to when using 100{\%} of labelled data in a conventional shallow network (i.e., increase in accuracy of 2.7{\%}, 3.7{\%}, 4.8{\%}, and 0.9{\%} across the activity recognition of four subjects). CONCLUSION: The approach proposed showed an improvement in the accuracy of classifying human activity when a limited amount of labelled data is available.",
keywords = "Activation function, Activity recognition, Data fusion, Neural networks",
author = "Netzahualcoyotl Hernandez-Cruz and CD Nugent and Ian McChesney and Shuai Zhang and Jesus Favela",
year = "2019",
month = "11",
day = "19",
doi = "http://dx.doi.org/10.4108/eai.30-10-2018.161669",
language = "English",
volume = "5",
journal = "EAI Endorsed Transactions on Pervasive Health and Technology",
issn = "2411-7145",
number = "17",

}

TY - JOUR

T1 - Adoption of the Activation Function Fusion Approach to Identify Human Activity Recognition in a Semi-Supervised Neural Network

AU - Hernandez-Cruz, Netzahualcoyotl

AU - Nugent, CD

AU - McChesney, Ian

AU - Zhang, Shuai

AU - Favela, Jesus

PY - 2019/11/19

Y1 - 2019/11/19

N2 - INTRODUCTION: Neural networks are a popular type of algorithm for human activity monitoring which can build intelligent systems from labelled data in an automated fashion. Obtaining accurately labelled data is costly; it requires time and effort, which can be cumbersome because it interrupts the user activity stream. In conjunction with the ubiquitous presence of embedded technology, neural networks present new research opportunities for human activity monitoring in smart home environments. OBJECTIVES: We propose a human activity classification method that requires a limited amount of labelled data, which consists of a concatenation method for classifying human activities built upon the fusion of neural network activation functions. METHODS: Our methodology builds a neural network model that receives the sensor data through the input layer to then distribute it among the different vertical hidden layers, which implement different activation functions simultaneously. Next a hidden layer combines activation functions by utilising a concatenation method. Finally, the neural network provides classes to the unlabelled sensing data. We conducted an evaluation utilising an open-access dataset. We compared the activity recognition accuracy of our approach utilising 25%, 50%, and 75% of labelled data against a conventional shallow neural network trained with the 100% of labelled data available. RESULTS: Results show an improvement in the accuracy of the activity classification regardless of the portion of labelled data available. It was observed that the highest achieved accuracy when using 25% of activation function fusion data outperformed results compared to when using 100% of labelled data in a conventional shallow network (i.e., increase in accuracy of 2.7%, 3.7%, 4.8%, and 0.9% across the activity recognition of four subjects). CONCLUSION: The approach proposed showed an improvement in the accuracy of classifying human activity when a limited amount of labelled data is available.

AB - INTRODUCTION: Neural networks are a popular type of algorithm for human activity monitoring which can build intelligent systems from labelled data in an automated fashion. Obtaining accurately labelled data is costly; it requires time and effort, which can be cumbersome because it interrupts the user activity stream. In conjunction with the ubiquitous presence of embedded technology, neural networks present new research opportunities for human activity monitoring in smart home environments. OBJECTIVES: We propose a human activity classification method that requires a limited amount of labelled data, which consists of a concatenation method for classifying human activities built upon the fusion of neural network activation functions. METHODS: Our methodology builds a neural network model that receives the sensor data through the input layer to then distribute it among the different vertical hidden layers, which implement different activation functions simultaneously. Next a hidden layer combines activation functions by utilising a concatenation method. Finally, the neural network provides classes to the unlabelled sensing data. We conducted an evaluation utilising an open-access dataset. We compared the activity recognition accuracy of our approach utilising 25%, 50%, and 75% of labelled data against a conventional shallow neural network trained with the 100% of labelled data available. RESULTS: Results show an improvement in the accuracy of the activity classification regardless of the portion of labelled data available. It was observed that the highest achieved accuracy when using 25% of activation function fusion data outperformed results compared to when using 100% of labelled data in a conventional shallow network (i.e., increase in accuracy of 2.7%, 3.7%, 4.8%, and 0.9% across the activity recognition of four subjects). CONCLUSION: The approach proposed showed an improvement in the accuracy of classifying human activity when a limited amount of labelled data is available.

KW - Activation function

KW - Activity recognition

KW - Data fusion

KW - Neural networks

UR - http://www.scopus.com/inward/record.url?scp=85076925988&partnerID=8YFLogxK

U2 - http://dx.doi.org/10.4108/eai.30-10-2018.161669

DO - http://dx.doi.org/10.4108/eai.30-10-2018.161669

M3 - Article

VL - 5

JO - EAI Endorsed Transactions on Pervasive Health and Technology

T2 - EAI Endorsed Transactions on Pervasive Health and Technology

JF - EAI Endorsed Transactions on Pervasive Health and Technology

SN - 2411-7145

IS - 17

M1 - e3

ER -