Steven McDonagh
YOU?
Author Swipe
View article: Rethinking Inter-LoRA Orthogonality in Adapter Merging: Insights from Orthogonal Monte Carlo Dropout
Rethinking Inter-LoRA Orthogonality in Adapter Merging: Insights from Orthogonal Monte Carlo Dropout Open
We propose Orthogonal Monte Carlo Dropout, a mechanism that enforces strict orthogonality when combining sparse semantic vectors without extra time complexity. Low-Rank Adaptation (LoRA), a popular fine-tuning method for large models, typi…
View article: A Shift in Perspective on Causality in Domain Generalization
A Shift in Perspective on Causality in Domain Generalization Open
The promise that causal modelling can lead to robust AI generalization has been challenged in recent work on domain generalization (DG) benchmarks. We revisit the claims of the causality and DG literature, reconciling apparent contradictio…
View article: SWiFT: Soft-Mask Weight Fine-tuning for Bias Mitigation
SWiFT: Soft-Mask Weight Fine-tuning for Bias Mitigation Open
Recent studies have shown that Machine Learning (ML) models can exhibit bias in real-world scenarios, posing significant challenges in ethically sensitive domains such as healthcare. Such bias can negatively affect model fairness, model ge…
View article: No time to train! Training-Free Reference-Based Instance Segmentation
No time to train! Training-Free Reference-Based Instance Segmentation Open
The performance of image segmentation models has historically been constrained by the high cost of collecting large-scale annotated data. The Segment Anything Model (SAM) alleviates this original problem through a promptable, semantics-agn…
View article: Concept-based Adversarial Attack: a Probabilistic Perspective
Concept-based Adversarial Attack: a Probabilistic Perspective Open
We propose a concept-based adversarial attack framework that extends beyond single-image perturbations by adopting a probabilistic perspective. Rather than modifying a single image, our method operates on an entire concept -- represented b…
View article: CheXGenBench: A Unified Benchmark For Fidelity, Privacy and Utility of Synthetic Chest Radiographs
CheXGenBench: A Unified Benchmark For Fidelity, Privacy and Utility of Synthetic Chest Radiographs Open
We introduce CheXGenBench, a rigorous and multifaceted evaluation framework for synthetic chest radiograph generation that simultaneously assesses fidelity, privacy risks, and clinical utility across state-of-the-art text-to-image generati…
View article: Exploiting Mixture-of-Experts Redundancy Unlocks Multimodal Generative Abilities
Exploiting Mixture-of-Experts Redundancy Unlocks Multimodal Generative Abilities Open
In this work, we undertake the challenge of augmenting the existing generative capabilities of pre-trained text-only large language models (LLMs) with multi-modal generation capability while satisfying two core constraints: C1 preserving t…
View article: Unlocking the Potential of Weakly Labeled Data: A Co-Evolutionary Learning Framework for Abnormality Detection and Report Generation
Unlocking the Potential of Weakly Labeled Data: A Co-Evolutionary Learning Framework for Abnormality Detection and Report Generation Open
Anatomical abnormality detection and report generation of chest X-ray (CXR) are two essential tasks in clinical practice. The former aims at localizing and characterizing cardiopulmonary radiological findings in CXRs, while the latter summ…
View article: There is no SAMantics! Exploring SAM as a Backbone for Visual Understanding Tasks
There is no SAMantics! Exploring SAM as a Backbone for Visual Understanding Tasks Open
The Segment Anything Model (SAM) was originally designed for label-agnostic mask generation. Does this model also possess inherent semantic understanding, of value to broader visual tasks? In this work we follow a multi-staged approach tow…
View article: Improving Object Detection via Local-global Contrastive Learning
Improving Object Detection via Local-global Contrastive Learning Open
Visual domain gaps often impact object detection performance. Image-to-image translation can mitigate this effect, where contrastive approaches enable learning of the image-to-image mapping under unsupervised regimes. However, existing met…
View article: BMFT: Achieving Fairness via Bias-based Weight Masking Fine-tuning
BMFT: Achieving Fairness via Bias-based Weight Masking Fine-tuning Open
Developing models with robust group fairness properties is paramount, particularly in ethically sensitive domains such as medical diagnosis. Recent approaches to achieving fairness in machine learning require a substantial amount of traini…
View article: einspace: Searching for Neural Architectures from Fundamental Operations
einspace: Searching for Neural Architectures from Fundamental Operations Open
Neural architecture search (NAS) finds high performing networks for a given task. Yet the results of NAS are fairly prosaic; they did not e.g. create a shift from convolutional structures to transformers. This is not least because the sear…
View article: Erase to Enhance: Data-Efficient Machine Unlearning in MRI Reconstruction
Erase to Enhance: Data-Efficient Machine Unlearning in MRI Reconstruction Open
Machine unlearning is a promising paradigm for removing unwanted data samples from a trained model, towards ensuring compliance with privacy regulations and limiting harmful biases. Although unlearning has been shown in, e.g., classificati…
View article: MULAN: A Multi Layer Annotated Dataset for Controllable Text-to-Image Generation
MULAN: A Multi Layer Annotated Dataset for Controllable Text-to-Image Generation Open
Text-to-image generation has achieved astonishing results, yet precise spatial controllability and prompt fidelity remain highly challenging. This limitation is typically addressed through cumbersome prompt engineering, scene layout condit…
View article: Label-efficient object detection via region proposal network pre-training
Label-efficient object detection via region proposal network pre-training Open
Self-supervised pre-training, based on the pretext task of instance discrimination, has fueled the recent advance in label-efficient object detection. However, existing studies focus on pre-training only a feature extractor network to lear…
View article: Optimisation-Based Multi-Modal Semantic Image Editing
Optimisation-Based Multi-Modal Semantic Image Editing Open
Image editing affords increased control over the aesthetics and content of generated images. Pre-existing works focus predominantly on text-based instructions to achieve desired image modifications, which limit edit precision and accuracy.…
View article: Multi-task Learning with 3D-Aware Regularization
Multi-task Learning with 3D-Aware Regularization Open
Deep neural networks have become a standard building block for designing models that can perform multiple dense computer vision tasks such as depth estimation and semantic segmentation thanks to their ability to capture complex correlation…
View article: Learning to Name Classes for Vision and Language Models
Learning to Name Classes for Vision and Language Models Open
Large scale vision and language models can achieve impressive zero-shot recognition performance by mapping class specific text queries to image content. Two distinct challenges that remain however, are high sensitivity to the choice of han…
View article: Tunable Convolutions with Parametric Multi-Loss Optimization
Tunable Convolutions with Parametric Multi-Loss Optimization Open
Behavior of neural networks is irremediably determined by the specific loss and data used during training. However it is often desirable to tune the model at inference time based on external factors such as preferences of the user or dynam…
View article: Label-Efficient Object Detection via Region Proposal Network Pre-Training
Label-Efficient Object Detection via Region Proposal Network Pre-Training Open
Self-supervised pre-training, based on the pretext task of instance discrimination, has fueled the recent advance in label-efficient object detection. However, existing studies focus on pre-training only a feature extractor network to lear…
View article: Content-Diverse Comparisons improve IQA
Content-Diverse Comparisons improve IQA Open
Image quality assessment (IQA) forms a natural and often straightforward undertaking for humans, yet effective automation of the task remains highly challenging. Recent metrics from the deep learning community commonly compare image pairs …
View article: CLAD: A realistic Continual Learning benchmark for Autonomous Driving
CLAD: A realistic Continual Learning benchmark for Autonomous Driving Open
In this paper we describe the design and the ideas motivating a new Continual Learning benchmark for Autonomous Driving (CLAD), that focuses on the problems of object classification and object detection. The benchmark utilises SODA10M, a r…
View article: Residual Contrastive Learning for Image Reconstruction: Learning Transferable Representations from Noisy Images
Residual Contrastive Learning for Image Reconstruction: Learning Transferable Representations from Noisy Images Open
This paper is concerned with contrastive learning (CL) for low-level image restoration and enhancement tasks. We propose a new label-efficient learning paradigm based on residuals, residual contrastive learning (RCL), and derive an unsuper…
View article: Model-Based Image Signal Processors via Learnable Dictionaries
Model-Based Image Signal Processors via Learnable Dictionaries Open
Digital cameras transform sensor RAW readings into RGB images by means of their Image Signal Processor (ISP). Computational photography tasks such as image denoising and colour constancy are commonly performed in the RAW domain, in part du…
View article: Out-of-Distribution Detection with Class Ratio Estimation
Out-of-Distribution Detection with Class Ratio Estimation Open
Density-based Out-of-distribution (OOD) detection has recently been shown unreliable for the task of detecting OOD images. Various density ratio based approaches achieve good empirical performance, however methods typically lack a principl…
View article: Re-examining Distillation For Continual Object Detection
Re-examining Distillation For Continual Object Detection Open
Training models continually to detect and classify objects, from new classes and new domains, remains an open problem. In this work, we conduct a thorough analysis of why and how object detection models forget catastrophically. We focus on…
View article: CroMo: Cross-Modal Learning for Monocular Depth Estimation
CroMo: Cross-Modal Learning for Monocular Depth Estimation Open
Learning-based depth estimation has witnessed recent progress in multiple directions; from self-supervision using monocular video to supervised methods offering highest accuracy. Complementary to supervision, further boosts to performance …
View article: Long-tail Recognition via Compositional Knowledge Transfer
Long-tail Recognition via Compositional Knowledge Transfer Open
In this work, we introduce a novel strategy for long-tail recognition that addresses the tail classes' few-shot problem via training-free knowledge transfer. Our objective is to transfer knowledge acquired from information-rich common clas…
View article: Spread Flows for Manifold Modelling
Spread Flows for Manifold Modelling Open
Flow-based models typically define a latent space with dimensionality identical to the observational space. In many problems, however, the data does not populate the full ambient data space that they natively reside in, rather inhabiting a…
View article: Flow Based Models For Manifold Data.
Flow Based Models For Manifold Data. Open
Flow-based generative models typically define a latent space with dimensionality identical to the observational space. In many problems, however, the data does not populate the full ambient data-space that they natively reside in, rather i…