Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 18 of 18 for “"Label-efficient Learning"”.
-
LABEL EFFICIENT LEARNING BEYOND MANUAL ANNOTATIONS
The supervised learning paradigm paired with the capacity of Deep Neural Networks (DNNs) have significantly advanced the state-of-the-art performance for many computer vision tasks. However, enormous amount of labeled data is required for supervised learning to achieve good generalization …
-
Label-Efficient Learning For Scene Segmentation
… challenge in the large scaling of deep learning systems. This thesis addresses this critical challenge of label scarcity in training object segmentation models, a task often hindered by the need for extensive and costly manual annotation. Our research explores two complementary …
-
Label Efficient Learning for Semantic Segmentation
… intelligence applications rooted in deep learning have made substantial advancements. These methods, however, are often characterised as ’data-hungry’, requiring vast quantities of labelled data to ensure effective generalisation across a variety of real-world scenarios. The process of …
-
Label-efficient learning for medical images
<p>Despite the improvements deep-learning methods brought to a great variety of computer vision tasks, the requirement of large-scale annotated datasets hinders many of their real-world applications. In the management of prostate cancer, where segmentation and registration of lower-pelvic imaging …
-
Label-efficient learning paradigms for scalable robot perception
… forecasting, and collision avoidance. However, learning such representations typically relies on dense and expensive forms of supervision such as 3D occupancy grids, bird's-eye view maps, and 3D object trajectories. Obtaining such annotations across diverse environments and sensor configurations …
-
LABEL-EFFICIENT LEARNING: BUILDING AI MODELS WITH LIMITED DATA
Modern deep learning relies on large-scale datasets with high-quality labels, which are expensive and time-consuming to acquire. In many real-world scenarios only limited labeled data is available, which is insufficient for training well-generalized models. This thesis advances label-efficient …
-
Modelling Human Behaviour in Smart Buildings Through Explainable Label-Efficient Learning Methods
… a central role in enabling data-driven, energy efficient control strategies that enhance sustainability while maintaining comfort. However, existing approaches often rely on large, labeled datasets collected from a single environment. When deployed in new settings, their performance degrades due …
-
Enhancing Microsleep Prediction Performance Through Temporal Modeling, Multimodal Fusion, and Label-Efficient Learning
… the development of robust EEG–EOG-based machine learning and deep learning frameworks for multimodal microsleep prediction, with a focus on physiologically meaningful representation learning, temporal modelling, multimodal interaction analysis, and practical real-world applicability. This thesis …
-
Deep Learning for 3D Point Clouds: Multimodal and Label-Efficient Learning for Geoscience Applications
… scarcity of annotated 3D datasets for supervised learning. To address these challenges, this dissertation develops a series of multimodal and data-efficient deep learning frameworks across diverse geoscientific contexts. Current approaches to multimodal fusion often convert 3D point clouds into 2D …
-
Resource-Aware and Label-Efficient Learning for Smart Buildings: Deployable Energy Disaggregation and Few-Shot Thermal Occupancy Estimation
… (NILM). Both are limited by the same realities: labelled data is scarce, sensor modalities differ across deployments, and inference targets are low-cost edge devices with only a few hundred kilobytes of memory. This thesis addresses these constraints through two complementary label-efficient and …
-
Label Efficient Learning for Medical Image Segmentation over Partially Labeled Dataset: Towards a Single, Unified and Lifelong Whole-body Organ Segmentation Model
Deep Learning has been widely used in the mainstream medical image segmentation methods such as tumor/lesion segmentation and organ-at-risk segmentation. Although deep learning models achieve leading performance in various medical segmentation tasks, training such a model requires large amounts of …
-
Efficient Visual Learning for Scene Understanding
… have been driven by deep neural networks. These learning-based frameworks enhance performance through extensive training datasets and a large number of trainable parameters. However, they are less scalable and require substantial computational and financial resources. This dissertation …
-
Learning Accurate and Robust Deep Visual Models
… We mainly focus on two questions: How to design label-efficient learning algorithms for computer vision tasks? How to improve the robustness of DNN based visual models? Concerning label-efficiency, we investigate a reinforced sequential model for video summarization, a background hallucination …
-
Learning to understand large-scale 3D point clouds
… semantic understanding to weakly supervised label efficient learning of large-scale 3D point clouds.</p> <p>The main contributions of this thesis are three-fold. In Chapter 3, we start by building an urban-scale photogrammetric point cloud dataset with nearly three billion richly annotated …
-
Applications and Methods for Energy-based Models at Scale
… to improve downstream discriminative machine learning applications. Generative models present an avenue for learning about underlying low-dimensional structure hidden within high-dimensional datasets and they can be trained on unlabeled data, enabling a pathway for building more …
-
Label-Efficient Segmentation for Diverse Scenarios
… In this thesis, we analyze the importance label-efficient segmentation techniques and provide a series of methods to address segmentation tasks in different scenarios. First, we propose a Deep Reasoning Network for few-shot semantic segmentation, termed as DRNet, which is a novel approach …
-
Minimizing label dependence in artificial intelligence for cancer surgery
… constrained by the scarcity of high-quality labelled data, particularly for emerging medical modalities where expert annotation is resource-intensive or infeasible. This thesis investigates label-efficient learning strategies to overcome this bottleneck, exploring two distinct paradigms: …