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 20 of 21 for “"Class-incremental Learning"”.
-
Recoverable Memory Bank For Class-Incremental Learning
Incremental learning aims to enable machine learning systems to sequentially learn new tasks without forgetting the old ones. While some existing methods, such as data replay-based and parameter isolation-based approaches, achieve remarkable results in incremental learning, they often suffer from …
-
Class Incremental Learning in Deep Neural Networks
… the use of graphical processing units, deep learning systems have shown tremendous potential in achieving super-human performance in many computer vision tasks. However, deep learning models are not able to learn continuously in scenarios where the data distribution is non-stationary or …
-
Intrusion Detection in IoT Network using Few-Shot Class Incremental Learning
… samples are available from the known attack classes in the initial stage of an intrusion detection system, classic network intrusion detection methods can deliver good performance. However, the learned knowledge is no longer suitable for new types of attacks with just a few samples. On the …
-
Referencing Unlabelled World Data to Prevent Catastrophic Forgetting in Class-incremental Learning
… of "catastrophic forgetting" in deep continual-learning systems. The term refers to severe performance degradation for older tasks, as a system learns new tasks that are presented sequentially. Most previous techniques have emphasized preservation of existing knowledge while learning new tasks, …
-
Advanced continual machine learning
Deep learning suffers from catastrophic forgetting when learning a dynamic environment, where the data is continuously gathered. This thesis is proposed to tackle the catastrophic forgetting problem under various settings and constraints, i.e., rehearsal-based, rehearsal-free, data scarcity, …
-
Incremental learning and federated learning for heterogeneous medical image analysis
Standard deep learning paradigm may not be practical over real-world heterogeneous medical data, where new disease merges over time with data acquired in a distributed manner across various hospitals. There have been approaches to facilitate the training of deep models over two primary categories …
-
Continual Learning of Object Classification in the Real World
Technological advances in deep learning have brought remarkable performance in the object classification task but only when all the training data of classes to be learned are available at the same time. However, real-world data continually evolve through time, resulting in ever-changing learning …
-
MODEL-CENTRIC COLLABORATIVE MACHINE LEARNING: SHARING, FUSION, MAINTENANCE AND APPLICATION
Federated learning (FL) enables collaborative model training across devices without sharing raw data, promoting privacy-preserving machine learning. This dissertation focuses on model-centric collaborative learning, which exchanges models rather than gradients to enhance privacy and …
-
Efficient Few-Shot Incremental Learning
Continual learning in data-scarce environments requires models to adapt to novel concepts without forgetting prior knowledge, a challenge exacerbated by strict computational constraints in real-world deployments. This thesis addresses the stability-plasticity-efficiency trilemma in both Few-Shot …
-
Delving deep into fetal ultrasound video analysis
… upon by the recent proliferation of machine learning algorithms. This thesis explores facets of automated analysis of fetal ultrasound videos and frames for a breadth of clinical applications including view classification, anatomical structure detection, continually learning algorithms over …
-
Efficient Knowledge Transfer and Adaptation for Speech and Beyond
… to address the limitations of transfer learning in dynamically evolving audio and speech processing contexts, particularly through novel approaches for class-incremental learning, parameter-efficient adaptation, and multimodal modeling. First, we provide a comprehensive framework for …
-
Interpretable and Incremental Deep Learning for Honey Botanical Origin Authentication Using Hyperspectral Imaging
… imaging (HSI) together with advanced deep learning systems. Hyperspectral imaging stands out because it records both detailed spectral and spatial data. The technology excels at detecting minor variations that differentiate various honey types. However, it is not without its hurdles. The …
-
Biologically inspired neuromorphic algorithms for low power learning
… fresh neurons to facilitate continual learning. This architectural plasticity is largely overlooked as a technique in conventional machine learning and is inspired by biological adult neurogenesis.<br/><br/>To underpin our research, we provide a unifying definition and mathematical …
-
Novel Geometric Methods for Deep Learning
… is to demystify the black-box nature of deep learning models and enhance various aspects of DNNs, including uncertainty awareness, data efficiency, learning continuity, and more.(I) Inherent uncertainty during the learning process of DNNs can cast doubts on their predictions. Ensemble …
-
Leveraging neuro-inspired mechanisms for adaptive and efficient deep learning
… the adaptability and efficiency of modern deep learning systems. We propose three neuro-inspired methods that address key challenges in real-world machine learning: continual learning and data efficiency. First, we introduce NISPA, a method inspired by the brain's sparse dynamic connectivity and …
-
Towards efficient deep learning for computer vision
Deep learning models require significant resources to deploy, limiting their widespread adoption. The aim of this thesis is to address this issue by proposing methods to make deep learning models more efficient for training and deployment.<br><br>One important aspect of machine learning is the …
-
Continual Learning for Classification Tasks
The traditional machine learning models can only learn from stationary data stream and are very vulnerable to the changing environment. To encourage the machine learners to be more human-like and practical in the real-world, continual learning was put forward and attracts a surge of attention in …
-
Continual Learning for an Ever Evolving and Intelligent Malware Classification System
Malware classification poses unique challenges for continual learning (CL) systems, driven by the daily influx of new samples and the evolving nature of malware threats that exploit new vulnerabilities. Antivirus vendors encounter hundreds of thousands of unique software pieces daily, encompassing …
Page 1 of 2