˽·¿¾ãÀÖ²¿

Dr Simon Coupland

Job: Principal Lecturer

Faculty: Computing, Engineering and Media

School/department: School of Computer Science and Informatics

Research group(s): Centre for Computational Intelligence (CCI)

Address: ˽·¿¾ãÀÖ²¿, The Gateway, Leicester, LE1 9BH

T: +44 (0)116 207 8419

E: simonc@dmu.ac.uk

W: /cci

 

Personal profile

Simon Coupland is a researcher working in the area of type-2 fuzzy logic.  Simon has worked on the underpinning mathematics of the field making a number of important contributions.  He also works on practical problems in this area including control, decision making and computing with words.

Research group affiliations

Centre for Computational Intelligence

Publications and outputs


  • dc.title: Assessing the Provenance of Student Coursework dc.contributor.author: Coupland, Simon dc.description.abstract: The Higher Education sector is mobilising vast resources in its response to the use of Generative AI in student coursework. This response includes institutional policies, training for staff and students and AI detection tools. This paper is concerned with one aspect of this fast-moving area; the assessment of the provenance of a piece of written student coursework. The question of the provenance of student work is a surprisingly complex one, which, in truth can only ever be answered by the student themselves. As academics we must understand the difference between checking for plagiarism and generative AI use. When assessing a student's possible use of generative AI there is no ground truth for us to test against and this makes the detection of AI use a completely different problem to plagiarism detection. A range of AI detection tools are available, some of which have been adopted within the sector. Some of these tools have high detection rates, however, most suffer with false positive rates meaning institutions would be falsely accusing hundreds of students per year of committing academic offences. This paper explores a different approach to this problem which complements the use of AI detection tools. Rather than examining the work submitted by a student, the author examines the creation and editing of the that work over time. This gives an understanding how a piece of work was written, and most importantly how it has been edited. Inspecting a documents history requires that it is written on a cloud-based platform with version history enabled. The author has created a tool which sits on top of the cloud-based platform and integrates with the virtual learning environment. The tool records each time a student digitally touches their work, and the changes are recorded. The tool interface gives an overview for a cohort, with the ability to delve more deeply into an individual submission. The result is an easily accessible interactive history of a document during its development, giving some kind of provenance to that document. This history of construction and editing, shows how a piece of written work has been crafted over time, providing useful evidence of academic practice. Data on the points where students digitally touch their work can also be useful beyond questions of academic practice. The Author gives an example of using a data-driven approach to give formative feedback and discusses how data-driven approaches could become common in teaching practice.

  • dc.title: Sprinting into blocks: what computing, AI and gaming academics learned dc.contributor.author: Allman, Zoe; Coupland, Simon; Khuman, A. S.; Fahy, Conor; Attwood, Luke

  • dc.title: Developing and delivering in block: Reflections one year in dc.contributor.author: Allman, Zoe; Coupland, Simon; Attwood, Luke; Fahy, Conor; Hasshu, Salim; Khuman, A. S.; Shell, Jethro

  • dc.title: Authentic assessment supporting curriculum and delivery mode transformation dc.contributor.author: Allman, Zoe; Coupland, Simon; Fahy, Conor dc.description.abstract: ˽·¿¾ãÀÖ²¿ is embracing significant transformation as curriculum and delivery mode transitions into an intensive block model approach. The Computer Games Programming (CGP) team were particularly innovative in their approach (Jones, 2022), completely revisiting curriculum sequencing and assessment methods to facilitate the best learning journey for students, and responding to employer and sector skills needs. This presentation highlights two examples of authentic assessments emerging from university-wide transformation. The digital economy requires graduates equipped with a set of digital skills which are practice based. CGP had been moving away from traditional written exams towards large, in-depth coursework which students produce over a term or academic year. In this approach digital skills are implicitly assessed, for example using source control metadata to assess students’ capabilities with a specific tool chain. This requires examination of digital footprints over the term. Therefore, the model needed revisiting to facilitate block delivery and explicitly assess these skills through face-to-face practical assessments we call driving tests, replicating assessments that are commonplace in other disciplines (Snodgrass et al, 2014; Kent-Waters et al, 2018). The depth of student knowledge is examined with a professional conversation, replicating assessment methods used in teacher and lecturer training (Britt et al, 2001). A driving test involves a student sitting with a tutor whilst being asked to perform a number of sequential pre-scripted tasks. Students are marked on the breadth of tasks they complete and the manner in which they complete them. The student is given immediate and personalised verbal feedback and an overall mark. The student leaves a digital trail which is used for moderation. Professional conversations introduce further diversity in assessment in level 6. These conversations supplement a practical assessment component and assess descriptors which can be difficult to evaluate in more traditional formats, for example identifying emerging issues at the forefront of the subject, and systematically identifying personal learning needs. Preparatory, co-created conversations highlighted that current level 6 learners would value this ‘technical interview’ format as the conversation allow learners to naturally demonstrate their understanding of the subject without additional coursework documentation/production. Students value these approaches that facilitate authentic demonstration of practical skills with tutor support and instant verbal feedback. As these assessment methods embed, there is ongoing consideration of whether these should be time limited activities; our experience suggests it should not as to date students have required different amounts of time to complete the task whilst demonstrating competency.

  • dc.title: A fast and efficient semantic short text similarity metric dc.contributor.author: Shell, Jethro; Coupland, Simon; Croft, David; Brown, Stephen dc.description.abstract: The semantic comparison of short sections of text is an emerging aspect of Natural Language Processing (NLP). In this paper we present a novel Short Text Semantic Similarity (STSS) method, Lightweight Semantic Similarity (LSS), to address the issues that arise with sparse text representation. The proposed approach captures the semantic information contained when comparing text to process the similarity. The methodology combines semantic term similarities with a vector similarity method used within statistical analysis. A modification of the term vectors using synset similarity values addresses issues that are encountered with sparse text. LSS is shown to be comparable to current semantic similarity approaches, LSA and STASIS, whilst having a lower computational footprint.

  • dc.title: A Neural Network for Interpolating Light-Sources dc.contributor.author: Colreavy-Donnelly, S.; Kuhn, Stefan; Caraffini, Fabio; O'Connor, S.; Anastassi, Zacharias; Coupland, Simon dc.description.abstract: This study combines two novel deterministic methods with a Convolutional Neural Network to develop a machine learning method that is aware of directionality of light in images. The first method detects shadows in terrestrial images by using a sliding-window algorithm that extracts specific hue and value features in an image. The second method interpolates light-sources by utilising a line-algorithm, which detects the direction of light sources in the image. Both of these methods are single-image solutions and employ deterministic methods to calculate the values from the image alone, without the need for illumination-models. They extract real-time geometry from the light source in an image, rather than mapping an illumination-model onto the image, which are the only models used today. Finally, those outputs are used to train a Convolutional Neural Network. This displays greater accuracy than previous methods for shadow detection and can predict light source-direction and thus orientation accurately, which is a considerable innovation for an unsupervised CNN. It is significantly faster than the deterministic methods. We also present a reference dataset for the problem of shadow and light direction detection. © 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

  • dc.title: Real-Time 3D Head Pose Tracking Through 2.5D Constrained Local Models with Local Neural Fields dc.contributor.author: Ackland, Stephen; Chiclana, Francisco; Istance, Howell; Coupland, Simon dc.description.abstract: Tracking the head in a video stream is a common thread seen within computer vision literature, supplying the research community with a large number of challenging and interesting problems. Head pose estimation from monocular cameras is often considered an extended application after the face tracking task has already been performed. This often involves passing the resultant 2D data through a simpler algorithm that best fits the data to a static 3D model to determine the 3D pose estimate. This work describes the 2.5D Constrained Local Model, combining a deformable 3D shape point model with 2D texture information to provide direct estimation of the pose parameters, avoiding the need for additional optimization strategies. It achieves this through an analytical derivation of a Jacobian matrix describing how changes in the parameters of the model create changes in the shape within the image through a full-perspective camera model. In addition, the model has very low computational complexity and can run in real-time on modern mobile devices such as tablets and laptops. The Point Distribution Model of the face is built in a unique way, so as to minimize the effect of changes in facial expressions on the estimated head pose and hence make the solution more robust. Finally, the texture information is trained via Local Neural Fields (LNFs) a deep learning approach that utilizes small discriminative patches to exploit spatial relationships between the pixels and provide strong peaks at the optimal locations. dc.description: The file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.

  • dc.title: Type-2 Fuzzy Elliptic Membership Functions for Modeling Uncertainty dc.contributor.author: Kayacan, E; Sarabakha, A; Coupland, Simon; John, Robert, 1955-; Ahmadieh, M.; Khanesar, M. A. dc.description.abstract: Whereas type-1 and type-2 membership functions (MFs) are the core of any fuzzy logic system, there are no performance criteria available to evaluate the goodness or correctness of the fuzzy MFs. In this paper, we make extensive analysis in terms of the capability of type-2 elliptic fuzzy MFs in modeling uncertainty. Having decoupled parameters for its support and width, elliptic MFs are unique amongst existing type-2 fuzzy MFs. In this investigation, the uncertainty distribution along the elliptic MF support is studied, and a detailed analysis is given to compare and contrast its performance with existing type-2 fuzzy MFs. Furthermore, fuzzy arithmetic operations are also investigated, and our finding is that the elliptic MF has similar features to the Gaussian and triangular MFs in addition and multiplication operations. Moreover, we have tested the prediction capability of elliptic MFs using interval type-2 fuzzy logic systems on oil price prediction problem for a data set from 2nd Jan 1985 till 25th April 2016. Throughout the simulation studies, an extreme learning machine is used to train the interval type-2 fuzzy logic system. The prediction results show that, in addition to their various advantages mentioned above, elliptic MFs have comparable prediction results when compared to Gaussian and triangular MFs. Finally, in order to test the performance of fuzzy logic controller with elliptic interval type-2 MFs, extensive real-time experiments are conducted for the 3D trajectory tracking problem of a quadrotor. We believe that the results of this study will open the doors to elliptic MFs’ wider use of real-world identification and control applications as the proposed MF is easy to interpret in addition to its unique features.

  • dc.title: Interval Type-2 Defuzzification Using Uncertainty Weights dc.contributor.author: Coupland, Simon; Runkler, Thomas; John, Robert, 1955-; Chen, Chao dc.description.abstract: One of the most popular interval type–2 defuzzification methods is the Karnik–Mendel (KM) algorithm. Nie and Tan (NT) have proposed an approximation of the KM method that converts the interval type–2 membership functions to a single type–1 membership function by averaging the upper and lower memberships, and then applies a type–1 centroid defuzzification. In this paper we propose a modification of the NT algorithm which takes into account the uncertainty of the (interval type–2) memberships. We call this method the uncertainty weight (UW) method. Extensive numerical experiments motivated by typical fuzzy controller scenarios compare the KM, NT, and UW methods. The experiments show that (i) in many cases NT can be considered a good approximation of KM with much lower computational complexity, but not for highly unbalanced uncertainties, and (ii) UW yields more reasonable results than KM and NT if more certain decision alternatives should obtain a larger weight than more uncertain alternatives.

  • dc.title: INNATE: Intelligent Non-invasive Nocturnal epilepsy Assistive TEchnology dc.contributor.author: Malekmohamadi, Hossein; Shell, Jethro; Coupland, Simon dc.description.abstract: Epilepsy is a neurological disease that affects the brain and is characterised by repeated seizures. Generalised, focal and unknown are three major types of seizures. Each type has several subgroups. For this reason, seizure detection and classification are expensive and erroneous. Other factors can also affect the detection. For example, patients can have a combination of different seizures or start with one type and finish with another. Nocturnal epilepsy can be prominent in many sufferers of this disease. This displays seizures that occur during the sleep cycle. The nature of such seizures makes the gathering of data and the subsequent detection and classification complex and costly. The current standard for seizure detection is the invasive use of electroencephalogram (EEG) monitoring. Both medical and research communities have expressed a large interest in the detection and classification of seizures automatically and non-invasively. This project proposes the use of 3D computer vision and pattern recognition techniques to detect seizures non-invasively.

 

Research interests/expertise

Understanding the performance capabilities of type-2 fuzzy logic.

Improving the computational performance of type-2 fuzzy logic systems.

Assessing other extensions to type-1 fuzzy sets and systems such as triangular type-2 fuzzy sets and non-stationary fuzzy systems.

The application of type-2 fuzzy logic to real-world problems.

Areas of teaching

MSc Computing/IT/ISM Introduction to computer systems.

Occasional lectures to MSc CIR on fuzzy logic, neural networks and recent advances in research.   

Qualifications

PhD in Computer Science

BSc (Hons) Computing

Courses taught

IMAT3404 Mobile Robots

Honours and awards

Joint Winner IEEE CIS Pre-college Education subcommittee Video Competition, 2012.

IEEE Transactions on Fuzzy Systems Outstanding Paper Award, 2009.

British Computer Society Machine Intelligence Award Winner, 2008.

Membership of professional associations and societies

IEEE Member

Externally funded research grants information

FuzzyPhoto, AHRC, 01/11/12 – 31/10/14, CI, Stephen Brown.