51黑料社区

More

Dr Sarah Greenfield

Job: Research Fellow

Faculty: Computing, Engineering and Media

School/department: School of Computer Science and Informatics

Research group(s): The 51黑料社区 Interdisciplinary Group in Intelligent Transport Systems (DIGITS), Centre for Computational Intelligence (CCI)

Address: 51黑料社区, The Gateway, Leicester, LE1 9BH

T: +44 (0)116 250 6171

E: s.greenfield@dmu.ac.uk

W: /digits

 

Close all sections
Open all sections

Personal profile

Sarah Greenfield received the BA in Mathematics and Philosophy from London University in 1978. In 2005 she was awarded a distinction in the MSc IT degree from 51黑料社区, Leicester, UK, and in 2012 she was awarded a PhD at De Montfort's Centre for Computational Intelligence, working under the supervision of Prof Chiclana. Her enduring interest in logic and the philosophy of mathematics was reflected in her original choice of degree subject. Her MSc project was in the field of type-2 fuzzy logic, and her PhD studies continued this theme in her exploration of mathematical and philosophical aspects of type-2 fuzzy logic in relation to such topics as uncertainty modelling and defuzzification.   Since completing her studies she has widened her research interests to include complex fuzzy inferencing and computational intelligence in transport.

Research group affiliations

  

Publications and outputs


  • dc.title: Enhancing Object Segmentation via聽Few-Shot Learning with聽Limited Annotated Data dc.contributor.author: Garc铆a-Aguilar, Iv谩n; Jafri, Syed Ali Haider; Elizondo, David; Calder贸n, Saul; Greenfield, Sarah; Luque-Baena, Rafael M. dc.description.abstract: Significant advancements in machine learning in recent years have revolutionized multiple sectors. The Segment-Anything Model (SAM) is a notable example of state-of-the-art image segmentation. Despite claims of zero-shot generalization, SAM exhibits limitations in specific scenarios like medical mammography images. SAM generates three segmentation masks per image to address this and recommends selecting the one with the highest confidence score. However, this is not always the optimal choice. This paper introduces a system that extends SAM鈥檚 segmentation capabilities by automatically selecting the correct mask, leveraging few-shot learning methods and an Out-of-Distribution threshold strategy. Several backbones were subjected to experimentation, highlighting the relationship between the support set size and the model鈥檚 accuracy.

  • dc.title: A Fuzzy Prescreening Tool to聽Assist in聽the聽Diagnosis of聽High Functioning Individuals on聽the聽Autism Spectrum Who Present with聽Mental Health Comorbidities dc.contributor.author: Smith, Philip; Greenfield, Sarah dc.description.abstract: Autism Spectrum Disorder is a neurological developmental disorder that effects at least 1% of the population, the majority of cases are high functioning individuals who struggle to get positive diagnoses that are vital to obtain community support. In this study, we have created and tested a Fuzzy Inferencing System to support clinicians, psychologists, family members and relevant stake holders to increase the chances for high functioning individuals to get a referral for full assessment to determine an autism diagnosis.

  • dc.title: Out-of-Distribution Detection with Memory-Augmented Variational Autoencoder dc.contributor.author: Ataeiasad, Faezeh; Elizondo, David; Ram铆rez, Sa煤l Calder贸n; Greenfield, Sarah; Deka, Lipika dc.description.abstract: This paper proposes a novel method capable of both detecting OOD data and generating in-distribution data samples. To achieve this, a VAE model is adopted and augmented with a memory module, providing capacities for identifying OOD data and synthesising new in-distribution samples. The proposed VAE is trained on normal data and the memory stores prototypical patterns of the normal data distribution. At test time, the input is encoded by the VAE encoder; this encoding is used as a query to retrieve related memory items, which are then integrated with the input encoding and passed to the decoder for reconstruction. Normal samples reconstruct well and yield low reconstruction errors, while OOD inputs produce high reconstruction errors as their encodings get replaced by retrieved normal patterns. Prior works use memory modules for OOD detection with autoencoders, but this method leverages a VAE architecture to enable generation abilities. Experiments conducted with CIFAR-10 and MNIST datasets show that the memory-augmented VAE consistently outperforms the baseline, particularly where OOD data resembles normal patterns. This notable improvement is due to the enhanced latent space representation provided by the VAE. Overall, the memory-equipped VAE framework excels in identifying OOD and generating creative examples effectively. dc.description: open access article

  • dc.title: Towards Refined Autism Screening: A Fuzzy Logic Approach with a Focus on Subtle Diagnostic Challenges dc.contributor.author: Smith, Philip; Greenfield, Sarah dc.description.abstract: This study explores the creation and testing of a Fuzzy Inferencing System for automating preliminary referrals for autism diagnosis, utilizing membership functions aligned with the Autism Quotient 10-item questionnaire. Validated across three distinct datasets, the system demonstrated perfect accuracy in deterministic settings and an overall accuracy of 92.91% in a broad fuzzy dataset. The use of Fuzzy Logic reflects the complex and variable nature of autism diagnosis, suggesting its potential applicability in this field. While the system effectively categorized clear referral and non-referral scenarios, it faced challenges in accurately identifying cases requiring a second opinion. These results indicate the need for further refinement to enhance the efficiency and accuracy of preliminary autism screenings, pointing to future avenues for improving the system鈥檚 performance. The motivation behind this study is to address the diagnostic gap for high-functioning adults whose symptoms present in a more neurotypical manner. Many current deep learning approaches for diagnosing autism focus on quantitative datasets like fMRI and facial expressions, often overlooking behavioral traits. However, autism diagnosis still heavily relies on long histories and multi-stakeholder information from parents, teachers, doctors and behavioral experts. This research addresses the challenge of creating an automated system that can handle the nuances and variability inherent in ASD symptoms. The theoretical innovation lies in the novel application of Fuzzy Logic to interpret these subtle diagnostic indicators, providing a more systematic approach compared to traditional methods. By bridging the gap between subjective clinical evaluations and objective computational techniques, this study aims to enhance the preliminary screening process for ASD. dc.description: open access article

  • dc.title: The Stratic Defuzzifier for Discretised General Type-2 Fuzzy Sets dc.contributor.author: Greenfield, Sarah; Chiclana, Francisco dc.description.abstract: Stratification is a feature of the type-reduced set of the general type-2 fuzzy set, from which a new technique for general type-2 defuzzification, Stratic Defuzzification, may be derived. Existing defuzzification s