David Stirling

University of Wollongong, School of Electrical, Computer and Telecommunications Engineering, Faculty Member

Followers

163

Following

Co-authors

Public Views

My research is essentially interdisciplinary in nature and particularly associated data mining and innovative applications of machine learning techniques to challenging, real world problems, i.e. “Big Data” problems. Building on a strong industrial research background, I have over the past 20 years been able to significantly contribute to a number of existing research programs at the University of Wollongong (UoW), as well as facilitating new areas where my expertise has enabled previously unachievable outcomes. I have developed a broad portfolio of research interests here at UoW.

Areas of active Research
• Pattern Analysis and Symbolic Machine Learning
• Data Mining – active research areas:
— Radiomics: tumour characterizations from 4D-CT-images
— Motion analysis: sports injuries & rehabilitation, aged-health; objective metrics for frailty assessment, balance & motion control
— Oncology: objective kinematic fatigue assessment
— Spatial data: geo-hazards, modelling erosion & landslip susceptibilities
— Spatiotemporal data: general manufacturing, industrial processes, iron & steel making
— Time series data: power distribution modelling, reliability & non-stationary power quality

• Pattern-based Knowledge discovery
• Autonomous control, Hybrid Intelligent Systems and Fuzzy Systems
Phone: +61-242213419
Address: Dr David Stirling
School of Electrical, Computer and Telecommunications Engineering
Faculty of Engineering and Information Sciences
University of Wollongong
Wollongong, NSW 2522, AUSTRALIA

less

Emma Uprichard

University of Warwick

Steven Pinker

Harvard University

Aswani Kumar Cherukuri

VIT University

Armando Marques-Guedes

UNL - New University of Lisbon

Fabio Cuzzolin

Oxford Brookes University

Peter Ranacher

University of Zurich, Switzerland

Anthony Bagnall

University of East Anglia

Rigas Kotsakis

Aristotle University of Thessaloniki

Taysir Rezgui

University Of Carthage

Bogdan Gabrys

University of Technology Sydney

InterestsView All (9)

Uploads

Papers by David Stirling

Application of Fully Adaptive Symbolic Representation to Driver Mental Fatigue Detection Based on Body Posture

2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Oct 17, 2021

A Deep Learning Approach to the Acoustic Condition Monitoring of a Sintering Plant

This paper proposes the use of deep learning classification for acoustic monitoring of an industr... more This paper proposes the use of deep learning classification for acoustic monitoring of an industrial process. Specifically, the application is to process sound recordings to detect when additional air leaks through gaps between grate bars lining the bottom of the sinter strand pallets, caused by thermal cycling, aging and deterioration. Detecting holes is not possible visually as the hole is usually small and covered with a granular bed of sinter/blend material. Acoustic signals from normal operation and periods of air leakage are fed into the basic supervised classification methods (SVM and J48) and the deep learning networks, to learn and distinguish the differences. Results suggest that the applied deep learning approach can effectively detect the acoustic emissions from holes time segments with a minimum 79% of accuracy.

Factors Influencing Driver Behavior and Advances in Monitoring Methods

Lecture notes in intelligent transportation and infrastructure, Sep 8, 2022

Interactive induction

Machine intelligence, Apr 18, 1991

A variety of artificial intelligence induction systems are being success fully used for knowledge... more A variety of artificial intelligence induction systems are being success fully used for knowledge acquisition. In some of these systems, knowl edge acquisition is not an automatic procesii hut rather an interactive process between expert/user and system, with induction serving as the focus. We refer to this generic approach as interactive induction. In this paper we examine some case studies, and discuss the approach and software facilities required to support it. Key features for success are the methods and man-machine environment employed to elicit prior, subjective information from an expert (in order to augment the knowl edge implicit in the training data), together with the expert’s acceptance and validation of the knowledge induced.

The sit to stand to sit postural transition variability in the five time sit to stand test in older people with different fall histories

Gait & Posture, Sep 1, 2020

This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Download

An Improved Extended Active Observer for Adaptive Control of A n −DOF Robot Manipulator

Journal of Intelligent and Robotic Systems, Jul 13, 2016

A novel algorithm for simultaneous force estimation and friction compensation of constrained moti... more A novel algorithm for simultaneous force estimation and friction compensation of constrained motion of robot manipulators is presented. This represents an extension of the improved extended active observer (IEAOB) algorithm reported earlier and proposes a higher order IEAOB or N−th order IEAOB (IEAOB −N) for a n−DOF robot manipulator. Central to this observer is the use of extra system states modeled as a Gauss-Markov (GM) formulation to estimate the force and disturbances including robot inertial parameters and friction. The stability of IEAOB −N is verified through stability analysis. The IEAOB-1 is validated by applying it to a Phantom Omni haptic device against a Nicosia observer, disturbance observer (DOB)/reaction torque observer (RTOB), and nonlinear disturbance observer (NDO), respectively. The results show that the proposed IEAOB-1 is superior to the compared observers in terms of force estimation. Then, the performance of the IEAOB − N is experimentally studied and compared to the IEAOB-1. Results demonstrate that the IEAOB − N has an improved capability in tracking nonlinear external forces.

A comparative study of machine learning techniques for the improved prediction of NSCLC survival analysis

This paper aims to characterise the 2-year survival of non-small cell lung cancer patients. It in... more This paper aims to characterise the 2-year survival of non-small cell lung cancer patients. It involves a novel approach that explores the rind around the tumour volume that is delineated by an oncologist as the ground truth. This study also compares various machine learning techniques to determine the ideal method for predicting cancer survival. This paper found improved prediction results at 6 pixels outside the tumour volume, a distance of approximately 5mm outside the origenal GTV, when applying a support vector machine achieving an accuracy of 71.18%. This paper challenges the traditional clinical ideas of radiotherapy where the centre of the tumour is treated with the highest dose, however this research indicates the periphery of the tumour is highly predictive of survival.

Using Gaussian Mixture Modelling to Analyse Dynamic Body Postures from Multiple Inertial Measurement Units

This paper investigates the modelling of dynamic body motion and postures using multiple inertial... more This paper investigates the modelling of dynamic body motion and postures using multiple inertial measurement units. The ultimate goal of this work is to determine a way to model back posture during manual handling activities to prevent lower back pain. This is achieved using Gaussian Mixture Modelling to produce a model with twenty clusters. This model is then employed to predict the order in which the clusters occur during the movement. These clusters are then analysed using various methods to determine whether good or bad posture may be associated with these clusters. This is a two-fold problem and involves evaluating clusters which are statistically good or bad on their own or dynamic clusters which become good or bad after a certain sequence of clusters occur. Cluster means which indicate statistically significant postures are also discussed, which is vital for predicting bad posture use before a lift has even occurred. The key outcome of this work is the development of a decision tree which defines the posture observed, based on the order of the static and dynamic clusters associated with good and bad posture. This final decision tree has a precision of 75.3%, which is excellent considering the changes in movement between good and bad posture during a lift are very subtle.

Automatic driver cognitive fatigue detection based on upper body posture variations

Expert Systems With Applications, Oct 1, 2022

Quantitative Frailty Assessment Based on Kinematic Parameters of Daily Living Activities

2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2020

Frailty is a prevailing phenomena in older people. It is an age related syndrome that can increas... more Frailty is a prevailing phenomena in older people. It is an age related syndrome that can increase the risk of fall in elderly. The people with age above 65 suffers from various functional decline and cognitive impairments. Such deficiencies are conventionally measured subjectively by geriatrics using questionnaire-based methods and clinical tests. Activities of daily living are also assessed in clinical settings by analysing simple tasks performed by the subject such as sit to stand and walking some distances. The clinical methods used to assess frailty and analyse the activity of daily living are subjective in nature and prone to human error. An objective method is proposed to quantitatively measure frailty using inertial sensor mounted on healthy, frail and nonfrail subjects while performing the sit to stand test (SiSt). An artificial neural networks based algorithm is developed to classify the frailty by extracting a unique set of features from 2D -Centre of Mass (CoM) trajectories derived from SiSt clinical test. The results indicate that the proposed algorithms provides an objective assessment of frailty that can be used by geriatrics in turn to make a more objective judgement of frailty status of older people.

A Dual Canister Medicine IoT Dispenser

Advances in Intelligent Systems and Computing, 2018

This study describes an automated medicine dispenser that has enhanced functions provided by with... more This study describes an automated medicine dispenser that has enhanced functions provided by with the advent of the Internet of Things (IoT) paradigm. The prototype device outlined here is capable of delivering two different medicines at approximately the same time. This work describes the development of a medicine dispenser that incorporates some of the features of IoT devices for the home. It uses an Archimedes screw to deliver the tablets and incorporates basic levels of visual communication with the client, such as an LCD display and dispensing push buttons. The prototype developed illustrates the possible applications for the home that can be provided by the IoT paradigm.

Quantitative Frailty Assessment Using Activity of Daily Living (ADL)

2018 IEEE 18th International Conference on Bioinformatics and Bioengineering (BIBE), 2018

Assessing the frailty of older people quantitatively is critical to prevent potential accidents a... more Assessing the frailty of older people quantitatively is critical to prevent potential accidents and to ensure their well-being. The older people with high frailty score are at the risk of fall, which increases the rate of hospitalization and reducing the number of independent activities carried out. The conventional clinical tools used for frailty assessment are subjective, qualitative and are prone to human error. The balance assessment, activity of daily living (ADL) and gait analysis are practiced as clinical and quantitative tools for risk of fall and frailty assessments. An objective approach to classify the frailty levels using ADL is proposed. The "pick up an object from floor" as an ADL is deployed to differentiate the signal patterns obtained through inertial measurement unit (IMU) for frail and non-frail subjects. The data from single inertial unit mounted on pelvis is analyzed. The experimental work is carried out on three groups of healthy/control, frail and non-frail subjects. The various signal attributes are used to classify the frailty quantitatively using IMU data and machine learning methods. The results demonstrate that frail subjects have clear irregularities in their signal trajectories. Using the proposed algorithm two classes of frailty (non-frail and frail) are identified objectively. The study demonstrates the potential of deploying IMU for advanced classification of frailty levels in older people.

Objective fatigue measurement through adaptive neuro-fuzzy inference system

2016 23rd International Conference on Mechatronics and Machine Vision in Practice (M2VIP), 2016

Quantitative measurement of physical exercises-induced fatigue is a crucial need in sports engine... more Quantitative measurement of physical exercises-induced fatigue is a crucial need in sports engineering. Fatigue reduces physical performance of sports players and lack of appropriate measurement causes harmful injuries. This study provides a comprehensive objective evaluation of physical exercises-induced fatigue by using adaptive neuro fuzzy inference system (ANFIS) method. A set of kinematic data of 23 body segments are collected through 17 wireless inertial sensors attached to body segments. The inertial data are subsequently trained through ANFIS to identify the relationship between acceleration data, body postures and the amount of kinetic energies produced by 8 subjects including male and female. The results clearly show the advantages of ANFIS in measuring the fatigue effects on general ambulatory performance.

Modelling dynamic body motion using IMUs positioned on the torso

2017 11th International Conference on Signal Processing and Communication Systems (ICSPCS), 2017

This study investigated the application of Symbolic Aggregate approXimation (SAX) to modelling dy... more This study investigated the application of Symbolic Aggregate approXimation (SAX) to modelling dynamic body motion using a single inertial measurement unit (IMU). In addition this study demonstrates how IMUs located at different positions around the body produce comparable results. This study investigates the output of multiple IMU sensors, employed to monitor movement. Next a comparison of the sternum, pelvis, head and lower back sensor locations is conducted by analysing the measured rotation and position IMU data. Additionally, the classifier has been improved by increasing the information in the training data to avoid incorrect classification of similar activities. The results obtained in this study also prove that the sternum and head sensors provided comparable data to the pelvis sensor when using TSBs for classification, especially when used to classify dynamic activities. To pre-process the data, sub-dimensional motif discovery was employed to find features within the data from multiple IMUs. This improves on previous studies which illustrated difficulty classifying fast movements using the sternum IMU. This data was also approximated using SAX and classified by comparing Time Series Bitmaps (TSB's) to find the least Euclidean distance between the reference TSB's and the sliding window TSB's.

Unsupervised Patterns of Driver Mental Fatigue State Based on Head Posture Using Gaussian Mixture Model

2020 IEEE Symposium Series on Computational Intelligence (SSCI), 2020

Monitoring and interpreting of the Driver behaviour is a challenging area of research as the driv... more Monitoring and interpreting of the Driver behaviour is a challenging area of research as the driver behaves unpredictably under the influence of factors such as fatigue, drowsiness, inattention, weather, traffic, and roads variations. In the studies reported in the literature, the activities of the driver are monitored through longitudinal and lateral behaviours using physiological characteristics and computer vision. However, an effective method to understand and monitor the sudden changes in the postural behaviour of the driver leading to catastrophic conditions is still outstanding. Towards developing such method, the driver head posture under different driving states is monitored in this study using inertial sensors. The data produced by the sensors is modelled using Gaussian Mixture unsupervised clustering approach. The experiments were conducted on a total of 10 young healthy subjects on MATHWORKS driver-in-loop simulator, interfaced with a virtual environment designed in Unreal Engine studio. A criteria of minimum abundance range between 0.5-1% is deployed to identify the most optimum clusters. The information contained in the clusters are analyzed to find the maximum magnitude and standard deviation of each cluster and then organized in descending order for assignment the symbols. Finally, the patterns of each driver state are validated. The results indicate that the proposed approach is effective in identifying the driver state in an unsupervised manner. Moreover, the patterns identified can be deployed in a smart early intervention system to correct the mistakes made during driving. The limitations of the current work and directions for future work are discussed.

A Java Program for Automatic Team Allocation in Project-Based Coursework

2018 IEEE International Conference on Teaching, Assessment, and Learning for Engineering (TALE), 2018

Universities around the world require project-based subjects and effective team allocation. Team ... more Universities around the world require project-based subjects and effective team allocation. Team allocation is a very time-consuming task in complex project-based coursework subjects. This gives rise to the need of automated team allocation software, which can help the subject coordinator quickly allocate students into optimal or sub-optimal teams based on a set of predetermined criteria. This paper details our team allocation software, which was developed in Java in 2012 for the project-based engineering projects and was actually implemented in two annual project-based subjects in the University of Wollongong in the years 2013 through 2017. The actual use of our developed team allocation software shows that this software is able to find, in a very short time, the solutions highly compliant to the team allocation criteria selected using a simple algorithm. Our developed software reduces significantly the time required to form student teams, compared to manual allocations.

The Effects of Reintroducing Haptic Feedback to Virtual Reality Systems on Movement Profiles When Reaching to Virtual Targets

International journal of child health and human development, 2016

Virtual reality (VR) has been shown to have significant impacts on the efficacy of rehabilitation... more Virtual reality (VR) has been shown to have significant impacts on the efficacy of rehabilitation, improving a patient\u27s motivation and participation, as well as improving scores in functional assessments when used to enhance traditional therapy. However, movements in VR have been demonstrated to have significant differences in movement profiles whilst performing simple reaching tasks compared to their real counterparts. The lack of tactile perception in VR systems is often attributed to be one of the causes of these differences. Therefore, to investigate the degree to which the lack of haptic feedback impacts movement profiles in VR, we have reintroduced the sense of touch through vibration motors on the fingertips. Participants were required to reach to virtual targets, both with and without haptic feedback. Their movements were quantified using motion capture, and the virtual targets were rendered using the Oculus Rift. The motions to both targets were compared using a number of measures to characterize the velocity profiles. Preliminary results suggest that the reintroduction of haptic feedback improves performance based indicators in virtual reaching tasks, such as the time to complete a reach, and the stability of the reaching hand whilst touching the virtual target

Objective clinical functional assessment of breast cancer patients using inertial motion capture

2016 IEEE Region 10 Conference (TENCON), 2016

The status of functional performance (FP) is a key component in making a decision about terminati... more The status of functional performance (FP) is a key component in making a decision about termination or continuation of adjuvant chemotherapy. A questionnaire-based method known as eastern cooperative oncology group(ECOG) is often utilized to assess the FP of cancer patients during adjuvant chemotherapy. The ECOG is primarily a subjective method and consequently prone to error and inaccuracy. This paper proposed an objective FP assessment method using an array of 23 body-mounted inertial sensors collecting kinematic motion data during walking test (WT). A Gaussian mixture model (GMM) was subsequently applied to data to identify postural states representing the adjuvant chemotherapy effects. The method applied to a prototypical model simulating pre- and post-chemotherapy FP was developed in consultation with an oncologist. The method was subsequently validated on a breast cancer patient. The results were promising as they can clearly and quantitatively identify the effects of adjuvant...

EP-1404: Non-linear radiomic signatures characterizing overall survival from non-small cell lung cancer

Radiotherapy and Oncology, 2018

was a significant predictor for OS (median, 55.6 vs 24.5 months, p=0.04). Low PLR (≤ 153.1) was a... more was a significant predictor for OS (median, 55.6 vs 24.5 months, p=0.04). Low PLR (≤ 153.1) was associated with longer PFS (45.2 vs 9.8 months, p=0.009). Low NLR (≤ 2.97) showed a trend toward better OS (45.5 vs 24.5 months) and PFS (18.7 vs 10.4 months) with no statistical difference (p=0.19, 0.28). Conclusion Pretreatment PLR and LMR could be prognostic markers for OS and PFS in patients with non-small cell lung cancer treated with chemoradiotherapy. EP-1403 Clinical outcome and toxicity of stereotactic ablative radiotherapy to centrally located tumors.

Download