Document not found! Please try again

Genetic Algorithms and Artificial Neural Networks for ...

1 downloads 0 Views 903KB Size Report
Nov 24, 2017 - 3Department of Mechanical Engineering, University of Auckland ..... Faculty of Creative Art and Industry, University of Auckland. Fred. Furlan.
2017 AUT Mathematical Sciences Symposium

Genetic Algorithms and Artificial Neural Networks for Optimising User Control in Hand Prosthetic Devices Luca Parisi1,2, Narrendar RaviChandran2,3 1Auckland

Bioengineering Institute (ABI), University of Auckland [email protected] 2University of Auckland Rehabilitative Technologies Association (UARTA), University of Auckland 3Department of Mechanical Engineering, University of Auckland Hand prosthetic devices have advanced rapidly in the last decades; however, for most people, such stateof-the-art prostheses are either unaffordable or too sophisticated to be used daily. Furthermore, current user control heavily relies on neuro-controlled mechanisms that contribute to increase the costs of such hand prostheses, further impairing a wide adoption and societal benefit of this technology. In this study, genetic algorithms were deployed for optimisation to select the appropriate set of learning parameters to use in a computationally efficient artificial neural network-based architecture used for classifying between cylindrical and tip grasps. The model was developed and tested in MATLAB (version 2017b, The MathWorks, Inc.). Electromyography signals on muscle activation recorded by the inexpensive MyoWare TM Muscle Sensor technology were used for model development, validation and testing via a 55-25-20 k-fold cross-validation algorithm. Cross validation is essential to avoid overfit-ting/overtraining, thus ensuring that the artificial neural network was truly learning from the data and, therefore, yielded reliable predictions. The model could classify between cylindrical and tip grasps in less than 1 second, with 89.2% (95% CI: 82.19% to 94.10%) classification accuracy, thus improving the performance of previous learning-based classifiers for the same task (87.12%±4.47%). Noteworthily, 100% classification accuracy was achieved in the training phase. Furthermore, the model achieved high sensitivity (95.00%, 95% CI: 86.08% to 98.96%) and specificity (83.33%, 95% CI: 71.48% to 91.71%) too. Once embedded onto an affordable prosthetic hand, these mathematical algorithms can enable upper extremity amputees to achieve agile, precise hand function/control in a very affordable way.

16

2017 AUT Mathematical Sciences Symposium Auckland University of Technology Auckland, New Zealand 23rd – 24th November 2017

Published by: Mathematical Sciences Research Group School of Engineering, Computer and Mathematical Sciences Auckland University of Technology http://www.aut.ac.nz/study-at-aut/study-areas/computer-mathematical-sciences/research-groups/ mathematical-sciences-research-group

Welcome to the 2017 AUT Mathematical Sciences Symposium On behalf of the Mathematical Sciences Research Group within the School of Engineering, Computer and Mathematical Sciences at Auckland University of Technology, we have much pleasure in welcoming you to the 2017 AUT Mathematical Sciences Symposium. This is the fourth such Symposium and it is a continuation of our efforts to develop and promote the research being undertaken within the Department of Mathematical Sciences as part of our recently enlarged School. We are delighted to welcome a number of invited speakers to the Symposium with the aim of exploring collaborative opportunities and potential new areas of research that can be established with our research active staff. The concept of this Symposium was a joint effort of us. We both appreciate the assistance of staff of the Department, in particular Dr Sarah Marshall, Dr Nuttanan Wichitaksorn and Dr Wenjun Zhang, who have each been involved in a variety of activities to ensure the continued success of this series. As New Zealand’s newest university we have recently had the opportunity to employ a number of new academic staff, all of whom have been developing research profiles. The School is putting in place a number of opportunities that will support and assist them in extending and enhancing their activities, with this meeting being one such effort. Our growing postgraduate programme in the Mathematical Sciences at Honours, Masters and Doctoral levels has been enhanced with our Master of Analytics (MAnalytics) degree, now in its third year. The success of this programme, with around thirty students at various stages of completion of the degree is leading to increased project supervision demands on our staff as well as leading to growing links with business and industry. We have established an arrangement with the SAS Institute that sees students in our MAnalytics degree gaining SAS Certification on graduation. The Mathematical Sciences Research Group focuses on two main areas – Analytics and Applied Mathematics. We are very much focused on “research lead teaching” and we have developed a small number of research clusters within these areas to strengthen and support those academic staff working in these areas. Ideally we would like to foster collaborative activities and we thank those of you who have joined us at this meeting and we hope that we can facilitate some future joint research efforts. We have kept the focus narrow so as to make the meeting meaningful and rewarding for those who participate. We hope that you enjoy your time with us and that you find the exercise a useful adjunct to the mathematical and statistical scene within New Zealand. On behalf of the Mathematical Sciences Research Group Jeffrey Hunter Professor of Mathematical Sciences Jiling Cao Professor of Mathematics Co-chairs of the 2017 AUT Mathematical Sciences Symposium

4

Symposium Schedule 8:30-9:00

Thursday 23rd November Registration

Friday 24th November

WF Level 7

Welcome

9:00-9:15

WF711

Matt Wand

Ilze Ziedins 9:15-10:00

9:00-9:45

A linear network with feedback and controls

Fast Approximate Inference for Arbitrarily Large Statistical Models via Message Passing

WF711

10:05 - 10:30

WF711

Murray Jorgensen

Hyuck Chung

WF711

WF514

9:50-10:15

10:30 - 11:00

Morning Tea

11:00 - 11:45

Preparing Financial Regulation for Forthcoming Crises

John Butcher

Samin Aref (Bibliometrix)

WF711

WF710

10:15-10:45

Morning Tea

10:45-11:30

Transport Equilibrium Models and Novel Applications

Andrea Raith

Eckhard Platen

WF711

11:50 - 12:15 12:15 - 12:40

WF711

Jose Da Fonseca

Catherine Hassell Sweatman

WF711

WF514

Jeong-Hoon Kim

Amir Rastar

WF711

WF514

12:00-12:25

Lunch

12:40 - 13:45

11:35-12:00

13:45-14:10

Correlated failures in multicomponent systems WF711

14:35-15:00 15:00-15:25

14:10-14:35

Winston Sweatman

Mohsen Pashna

WF711

WF710

Graeme Wake

Luca Parisi (Genetic Alg)

WF711

WF710

WF711

WF710

16:10-16:45

Nuttanan Wichitaksorn

Md. Shahidul Islam

Robin Willink

(Garment Industry) WF710

Marco Reale Parsimonious structures for multivariate time series models WF711

18:30 - 21:00

14:40-15:25 15:25-15:40

15:45-16:10

16:50-17:35

Sarah Marshall

Samin Aref (Binary Prog)

WF711

WF710

Lunch Jason Chen

Luca Parisi (AI)

WF711

WF710

Narrendar RaviChandran

Md. Shahidul Islam

WF711

WF710

(Demand Forecasting)

Which mechanisms shape species' responses to environmental problems like climate change WF711

Wenjun Zhang

WF711

WF710

William Godsoe

Afternoon Tea

15:25-15:45

Valerie Chopovda

WF711

12:25-13:45

Richard Arnold 13:45-14:30

Robin Hankin

Symposium Dinner Four Seasons Restaurant, WH Building Cnr Mayoral Drive and Wellesley Street East

6

Farewell WF711

List of Authors and Titles Samin Aref, Andrew J. Mason, Mark C. Wilson Computing the frustration index in signed graphs using binary programming . . . . . . . 10 Samin Aref, David Friggens, Shaun Hendy Analysing Scientific Collaborations of New Zealand Institutions using Scopus Bibliometric Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Richard Arnold, Stefanka Chukova, Yu Hayakawa Correlated failures in multicomponent systems . . . . . . . . . . . . . . . . . . . . . . 11 John Butcher Polish notation and numerical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Jason Chen A Spatiotemporal Modelling Study of a GnRH Neuron . . . . . . . . . . . . . . . . . . 11 Valerie Chopovda, Anton Gulley, Winston Sweatman Using magnets for particle extraction from powder flow . . . . . . . . . . . . . . . . . . 12 Hyuck Chung Acoustic pressure fields of 2D elastic cylindrical shells . . . . . . . . . . . . . . . . . . 12 Jose Da Fonseca, Edem Dawui Semivariance and Semiskew Risk Premiums in Currency Markets . . . . . . . . . . . . 12 William Godsoe Which mechanisms shape species’ responses to environmental problems like climate change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Robin Hankin Generalized Bradley-Terry models for MasterChef Australia . . . . . . . . . . . . . . . 13 Catherine Hassell Sweatman Mathematical model of diabetes and lipid metabolism : predicting the consequences of a low carbohydrate healthy fat diet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Md. Shahidul Islam, Jannat Ara Ferdousi Factors Influencing on Growth of Garment Industry . . . . . . . . . . . . . . . . . . . . 14 Md. Shahidul Islam, Mohammad Sazzad Mosharrof Comparison of Various Technique for Demand Forecasting in a Multinational Company . 14 Murray Jorgensen Clustering via Mixture Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Jeong-Hoon Kim SEV, Mellin Transform and Pricing of Financial Derivatives . . . . . . . . . . . . . . . 15 Sarah Marshall, Thomas Archibald Lot-sizing for a Product Recovery System with Quality-dependent Recovery Channels . 15 8

2017 AUT Mathematical Sciences Symposium

Luca Parisi, Marianne Lyne Manaog The Importance of Selecting Appropriate k-fold Cross-validation and Training Algorithms in Improving Postoperative Discharge Decision-making via Artificial Intelligence

16

Luca Parisi, Narrendar RaviChandran Genetic Algorithms and Artificial Neural Networks for Optimising User Control in Hand Prosthetic Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Mohsen Pashna, Rubiyah Yusof, Neda Afshar Oil Spill Dispersion Simulation, and Multi Robot System to track and predict the pollution trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Eckhard Platen, Michael Schmutz Preparing Financial Regulation for Forthcoming Crises . . . . . . . . . . . . . . . . . . 17 Andrea Raith Transport Equilibrium Models and Novel Applications . . . . . . . . . . . . . . . . . . 17 Amir Rastar Tracking for convective flow inside the respiratory airways using method of characteristics 18 Narrendar RaviChandran, Luca Parisi Bifurcation Analysis to Improve the Specificity of Biophysical Neuronal Models . . . . 18 Marco Reale Parsimonious structures for multivariate time series models . . . . . . . . . . . . . . . . 19 Winston Sweatman Memorable MISGs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Graeme Wake Asia-Pacific Initiatives for Mathematics-in-Industry . . . . . . . . . . . . . . . . . . . . 19 Matt Wand Fast Approximate Inference for Arbitrarily Large Statistical Models via Message Passing 20 Nuttanan Wichitaksorn, Boris Choy, Richard Gerlach Modeling Non-linear Dependence of Bivariate Seemingly Unrelated Tobit Models through Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Robin Willink The Guide to the Expression of Uncertainty in Measurement: a progressive retrospective

20

Wenjun Zhang, Jin. E. Zhang Changing Probability Measures in GARCH Option Pricing Models . . . . . . . . . . . . 21 Ilze Ziedins, Bill Helton, Frank Kelly, Ruth Williams A linear network with feedback and controls . . . . . . . . . . . . . . . . . . . . . . . . 21

9

List of Participants First name Samin Richard Boris Christian Murray John Jiling Jason Valerie Renu Hyuck Jose Neda Fred William Robin Catherine Jeffrey Gulshad Md. Shahidul

Surname Aref Arnold Bacic Bläsche Black Butcher Cao Chen Chopovda Choudhary Chung Da Fonseca Farrokhi Afshar Furlan Godsoe Hankin Hassell Sweatman Hunter Imran Islam

Arunkumar Murray Jeong-Hoon Sarah Julian Luca Mohsen

Jayakumar Jorgensen Kim Marshall Mccree Parisi Pashna

Balint

Petro

Eckhard

Platen

Andrea Amir Narrendar Marco Shu

Raith Rastar RaviChandran Reale Su

Winston Brice Alna Graeme Matt Maojun Stuart

Sweatman Valles van der Merwe Wake Wand Wang Weston

Nuttanan Robin Xiangjie Sidra Wenjun Ilze

Wichitaksorn Willink Xue Zafar Zhang Ziedins

Affiliation Department of Computer Science, University of Auckland School of Mathematics and Statistics, Victoria University of Wellington Computer Science, Auckland University of Technology Institute of Natural and Mathematical Sciences, Massey University Department of Mathematical Sciences, Auckland University of Technology Mathematics, University of Auckland Department of Mathematical Sciences , Auckland University of Technology Department of Mathematical Science, Auckland University of Technology Institute of Natural and Mathematical Sciences, Massey University Department of Mathematical Sciences, Auckland University of Technology Department of Mathematical Sciences, Auckland University of Technology Department of Finance, Auckland University of Technology Faculty of Creative Art and Industry, University of Auckland Department of Mathematical Sciences, Auckland University of Technology Bioprotection research centre, Lincoln University Department of Mathematical Sciences, Auckland University of Technology Department of Mathematical Sciences, Auckland University of Technology Department of Mathematical Sciences, Auckland University of Technology Mathematical Sciences, Auckland University of Technology School of Engineering, Computer and Mathematical Science, Auckland University of Technology Department of Mechanical Engineering, Auckland University of Technology Department of Mathematical Sciences, AUT, Auckland University of Technology Department of Mathematics, Yonsei Uiversity Department of Mathematical Sciences, Auckland University of Technology Data analytics, Auckland University of Technology Auckland Bioengineering Institute (ABI), University of Auckland Malaysia Japan International Institute of Technology, Centre for Artificial and Intelligence and Robotics, UTM Visiting from Budapest University of Technology and Economics, Auckland University of Technology School of Mathematical and Physical Sciences, Finance Discipline Group, University of Technology Sydney Department of Engineering Science, University of Auckland Auckland Bioengineering Institute, University of Auckland Department of Mechanical Engineering, University of Auckland School of Mathematics and Statistics, University of Canterbury School of Engineering, Computer and Mathematical Sciences, Auckland University of Technology Institute of Natural and Mathematical Sciences, Massey University Wholesale Analytics, Genesis Department of Mathematical Sciences, Auckland University of Technology Institute of Natural and Mathematical Sciences, Massey University School of Mathematical and Physical Sciences, University of Technology Sydney Department of Mathematical Sciences, Auckland University of Technology Institute for Radio Astronomy and Space Research , Auckland University of Technology Department of Mathematical Sciences, Auckland University of Technology Ministry of Business, Innovation and Employment Department of Statistics, University of Auckland Institute of Natural and Mathematical Sciences, Massey University Department of Mathematical Sciences, Auckland University of Technology Department of Statistics, University of Auckland

23

Suggest Documents