All Outputs (43)

Sparse Autoencoders Do Not Find Canonical Units of Analysis (2025)
Presentation / Conference Contribution
Leask, P., Bussmann, B., Pearce, M. T., Isaac Bloom, J., Tigges, C., Al Moubayed, N., Sharkey, L., & Nanda, N. (2025, April). Sparse Autoencoders Do Not Find Canonical Units of Analysis. Presented at The Thirteenth International Conference on Learning Representations, Singapore

A common goal of mechanistic interpretability is to decompose the activations of neural networks into features: interpretable properties of the input computed by the model. Sparse autoencoders (SAEs) are a popular method for finding these features in... Read More about Sparse Autoencoders Do Not Find Canonical Units of Analysis.

Disentangling Racial Phenotypes: Fine-Grained Control of Race-related Facial Phenotype Characteristics (2024)
Presentation / Conference Contribution
Yucer, S., Abarghouei, A. A., Al Moubayed, N., & Breckon, T. P. (2024, June). Disentangling Racial Phenotypes: Fine-Grained Control of Race-related Facial Phenotype Characteristics. Presented at 2024 International Joint Conference on Neural Networks (IJCNN), Yokohama, Japan

Achieving an effective fine-grained appearance variation over 2D facial images, whilst preserving facial identity, is a challenging task due to the high complexity and entanglement of common 2D facial feature encoding spaces. Despite these challenges... Read More about Disentangling Racial Phenotypes: Fine-Grained Control of Race-related Facial Phenotype Characteristics.

CXR-IRGen: An Integrated Vision and Language Model for the Generation of Clinically Accurate Chest X-Ray Image-Report Pairs (2024)
Presentation / Conference Contribution
Shentu, J., & Al Moubayed, N. (2024, January). CXR-IRGen: An Integrated Vision and Language Model for the Generation of Clinically Accurate Chest X-Ray Image-Report Pairs. Presented at 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Waikoloa, Hawaii, USA

Chest X-Ray (CXR) images play a crucial role in clinical practice, providing vital support for diagnosis and treatment. Augmenting the CXR dataset with synthetically generated CXR images annotated with radiology reports can enhance the performance of... Read More about CXR-IRGen: An Integrated Vision and Language Model for the Generation of Clinically Accurate Chest X-Ray Image-Report Pairs.

Length is a Curse and a Blessing for Document-level Semantics (2023)
Presentation / Conference Contribution
Xiao, C., Li, Y., Hudson, G. T., Lin, C., & Al Moubayed, N. (2023, December). Length is a Curse and a Blessing for Document-level Semantics. Presented at The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP), Singapore

In recent years, contrastive learning (CL) has been extensively utilized to recover sentence and document-level encoding capability from pre-trained language models. In this work, we question the length generalizability of CL-based models, i.e., thei... Read More about Length is a Curse and a Blessing for Document-level Semantics.

On Isotropy, Contextualization and Learning Dynamics of Contrastive-based Sentence Representation Learning (2023)
Presentation / Conference Contribution
Xiao, C., Long, Y., & Al Moubayed, N. (2023, July). On Isotropy, Contextualization and Learning Dynamics of Contrastive-based Sentence Representation Learning. Presented at Findings of the Association for Computational Linguistics: ACL 2023, Toronto, Canada

Addressing Performance Inconsistency in Domain Generalization for Image Classification (2023)
Presentation / Conference Contribution
Stirling, J., & Moubayed, N. A. (2023, June). Addressing Performance Inconsistency in Domain Generalization for Image Classification. Presented at 2023 International Joint Conference on Neural Networks (IJCNN), Gold Coast, Australia

Domain Generalization (DG) in computer vision aims to replicate the human ability to generalize well under a shift of data distribution, or domain. In recent years, the field of domain generalization has seen a steady increase in average left-out tes... Read More about Addressing Performance Inconsistency in Domain Generalization for Image Classification.

Natural Language Explanations for Machine Learning Classification Decisions (2023)
Presentation / Conference Contribution
Burton, J., Al Moubayed, N., & Enshaei, A. (2023, June). Natural Language Explanations for Machine Learning Classification Decisions. Presented at 2023 International Joint Conference on Neural Networks (IJCNN), Gold Coast, Australia

This paper addresses the challenge of providing understandable explanations for machine learning classification decisions. To do this, we introduce a dataset of expert-written textual explanations paired with numerical explanations, forming a data-to... Read More about Natural Language Explanations for Machine Learning Classification Decisions.

Is Unimodal Bias Always Bad for Visual Question Answering? A Medical Domain Study with Dynamic Attention (2022)
Presentation / Conference Contribution
Sun, Z., Harit, A., Cristea, A. I., Yu, J., Al Moubayed, N., & Shi, L. (2022, December). Is Unimodal Bias Always Bad for Visual Question Answering? A Medical Domain Study with Dynamic Attention. Presented at IEEE Big Data, Osaka, Japan

Medical visual question answering (Med-VQA) is to answer medical questions based on clinical images provided. This field is still in its infancy due to the complexity of the trio formed of questions, multimodal features and expert knowledge. In this... Read More about Is Unimodal Bias Always Bad for Visual Question Answering? A Medical Domain Study with Dynamic Attention.

Towards Graph Representation Learning Based Surgical Workflow Anticipation (2022)
Presentation / Conference Contribution
Zhang, X., Al Moubayed, N., & Shum, H. P. (2022, September). Towards Graph Representation Learning Based Surgical Workflow Anticipation. Presented at 2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI), Ioannina, Greece

Surgical workflow anticipation can give predictions on what steps to conduct or what instruments to use next, which is an essential part of the computer-assisted intervention system for surgery, e.g. workflow reasoning in robotic surgery. However, cu... Read More about Towards Graph Representation Learning Based Surgical Workflow Anticipation.

Does lossy image compression affect racial bias within face recognition? (2022)
Presentation / Conference Contribution
Yucer, S., Poyser, M., Al Moubayed, N., & Breckon, T. (2022, October). Does lossy image compression affect racial bias within face recognition?. Presented at International Joint Conference on Biometrics (IJCB 2022), Abu Dhabi, UAE

This study investigates the impact of commonplace lossy image compression on face recognition algorithms with regard to the racial characteristics of the subject. We adopt a recently proposed racial phenotype-based bias analysis methodology to measur... Read More about Does lossy image compression affect racial bias within face recognition?.

INTERACTION: A Generative XAI Framework for Natural Language Inference Explanations (2022)
Presentation / Conference Contribution
Yu, J., Cristea, A. I., Harit, A., Sun, Z., Aduragba, O. T., Shi, L., & Al Moubayed, N. (2022, July). INTERACTION: A Generative XAI Framework for Natural Language Inference Explanations. Presented at 2022 International Joint Conference on Neural Networks (IJCNN), Padova, Italy

XAI with natural language processing aims to produce human-readable explanations as evidence for AI decisionmaking, which addresses explainability and transparency. However, from an HCI perspective, the current approaches only focus on delivering a s... Read More about INTERACTION: A Generative XAI Framework for Natural Language Inference Explanations.

Efficient Uncertainty Quantification for Multilabel Text Classification (2022)
Presentation / Conference Contribution
Yu, J., Cristea, A. I., Harit, A., Sun, Z., Aduragba, O. T., Shi, L., & Al Moubayed, N. (2022, July). Efficient Uncertainty Quantification for Multilabel Text Classification. Presented at 2022 International Joint Conference on Neural Networks (IJCNN), Padova, Italy

Despite rapid advances of modern artificial intelligence (AI), there is a growing concern regarding its capacity to be explainable, transparent, and accountable. One crucial step towards such AI systems involves reliable and efficient uncertainty qua... Read More about Efficient Uncertainty Quantification for Multilabel Text Classification.

Contrastive Learning with Heterogeneous Graph Attention Networks on Short Text Classification (2022)
Presentation / Conference Contribution
Sun, Z., Harit, A., Cristea, A. I., Yu, J., Shi, L., & Al Moubayed, N. (2022, July). Contrastive Learning with Heterogeneous Graph Attention Networks on Short Text Classification. Presented at 2022 International Joint Conference on Neural Networks (IJCNN), Padova, Italy

Graph neural networks (GNNs) have attracted extensive interest in text classification tasks due to their expected superior performance in representation learning. However, most existing studies adopted the same semi-supervised learning setting as the... Read More about Contrastive Learning with Heterogeneous Graph Attention Networks on Short Text Classification.

MuLD: The Multitask Long Document Benchmark (2022)
Presentation / Conference Contribution
Hudson, G. T., & Al Moubayed, N. (2022, June). MuLD: The Multitask Long Document Benchmark. Presented at 13th Conference on Language Resources and Evaluation (LREC 2022), Marseille, France

The impressive progress in NLP techniques has been driven by the development of multi-task benchmarks such as GLUE and SuperGLUE. While these benchmarks focus on tasks for one or two input sentences, there has been exciting work in designing efficien... Read More about MuLD: The Multitask Long Document Benchmark.

Generating Textual Explanations for Machine Learning Models Performance: A Table-to-Text Task (2022)
Presentation / Conference Contribution
Ampomah, I., Burton, J., Enshaei, A., & Al Moubayed, N. (2022, June). Generating Textual Explanations for Machine Learning Models Performance: A Table-to-Text Task. Presented at 13th Conference on Language Resources and Evaluation (LREC 2022), Marseille, France

Numerical tables are widely employed to communicate or report the classification performance of machine learning (ML) models with respect to a set of evaluation metrics. For non-experts, domain knowledge is required to fully understand and interpret... Read More about Generating Textual Explanations for Machine Learning Models Performance: A Table-to-Text Task.

Enhanced Methods for Evolution in-Materio Processors (2022)
Presentation / Conference Contribution
Jones, B. A., Al Moubayed, N., Zeze, D. A., & Groves, C. (2023, November). Enhanced Methods for Evolution in-Materio Processors. Presented at IEEE International Conference on Rebooting Computing (ICRC 2021), Virtual

Evolution-in-Materio (EiM) is an unconventional computing paradigm, which uses an Evolutionary Algorithm (EA) to configure a material's parameters so that it can perform a computational task. While EiM processors show promise, slow manufacturing and... Read More about Enhanced Methods for Evolution in-Materio Processors.

Measuring Hidden Bias within Face Recognition via Racial Phenotypes (2022)
Presentation / Conference Contribution
Yucer, S., Tekras, F., Al Moubayed, N., & Breckon, T. (2022, January). Measuring Hidden Bias within Face Recognition via Racial Phenotypes. Presented at Proc. Winter Conference on Applications of Computer Vision, Waikoloa, HI

Recent work reports disparate performance for intersectional racial groups across face recognition tasks: face verification and identification. However, the definition of those racial groups has a significant impact on the underlying findings of such... Read More about Measuring Hidden Bias within Face Recognition via Racial Phenotypes.

Agree to Disagree: When Deep Learning Models With Identical Architectures Produce Distinct Explanations (2022)
Presentation / Conference Contribution
Watson, M., Awwad Shiekh Hasan, B., & Al Moubayed, N. (2022, January). Agree to Disagree: When Deep Learning Models With Identical Architectures Produce Distinct Explanations. Presented at Proc. Winter Conference on Applications of Computer Vision, Waikoloa, HI

Deep Learning of neural networks has progressively become more prominent in healthcare with models reaching, or even surpassing, expert accuracy levels. However, these success stories are tainted by concerning reports on the lack of model transparenc... Read More about Agree to Disagree: When Deep Learning Models With Identical Architectures Produce Distinct Explanations.

A Generative Bayesian Graph Attention Network for Semi-supervised Classification on Scarce Data (2021)
Presentation / Conference Contribution
Sun, Z., Harit, A., Yu, J., Cristea, A., & Al Moubayed, N. (2021, July). A Generative Bayesian Graph Attention Network for Semi-supervised Classification on Scarce Data. Presented at IEEE International Joint Conference on Neural Network (IJCNN2021), Virtual

This research focuses on semi-supervised classification tasks, specifically for graph-structured data under datascarce situations. It is known that the performance of conventional supervised graph convolutional models is mediocre at classification ta... Read More about A Generative Bayesian Graph Attention Network for Semi-supervised Classification on Scarce Data.

Towards Equal Gender Representation in the Annotations of Toxic Language Detection (2021)
Presentation / Conference Contribution
Excell, E., & Al Moubayed, N. (2021, August). Towards Equal Gender Representation in the Annotations of Toxic Language Detection. Presented at 3rd Workshop on Gender Bias in Natural Language Processing (GeBNLP2021), International Joint Conference on Natural Language Processing (INCNLP2021), Bangkok, Thailand

Classifiers tend to propagate biases present in the data on which they are trained. Hence, it is important to understand how the demographic identities of the annotators of comments affect the fairness of the resulting model. In this paper, we focus... Read More about Towards Equal Gender Representation in the Annotations of Toxic Language Detection.