Open Access. Powered by Scholars. Published by Universities.®
- Publication
Articles 1 - 4 of 4
Full-Text Articles in Engineering
Model-Free And Model-Based Active Learning For Regression, Jack O'Neill, Sarah Jane Delany, Brian Macnamee
Model-Free And Model-Based Active Learning For Regression, Jack O'Neill, Sarah Jane Delany, Brian Macnamee
Conference papers
Training machine learning models often requires large labelled datasets, which can be both expensive and time-consuming to obtain. Active learning aims to selectively choose which data is labelled in order to minimize the total number of labels required to train an effective model. This paper compares model-free and model-based approaches to active learning for regression, finding that model-free approaches, in addition to being less computationally intensive to implement, are more effective in improving the performance of linear regressions than model-based alternatives.
Activist: A New Framework For Dataset Labelling, Jack O'Neill, Sarah Jane Delany, Brian Macnamee
Activist: A New Framework For Dataset Labelling, Jack O'Neill, Sarah Jane Delany, Brian Macnamee
Conference papers
Acquiring labels for large datasets can be a costly and time-consuming process. This has motivated the development of the semi-supervised learning problem domain, which makes use of unlabelled data — in conjunction with a small amount of labelled data — to infer the correct labels of a partially labelled dataset. Active Learning is one of the most successful approaches to semi-supervised learning, and has been shown to reduce the cost and time taken to produce a fully labelled dataset. In this paper we present Activist; a free, online, state-of-the-art platform which leverages active learning techniques to improve the efficiency …
An Evaluation Of Selection Strategies For Active Learning With Regression, Jack O'Neill
An Evaluation Of Selection Strategies For Active Learning With Regression, Jack O'Neill
Dissertations
While active learning for classification problems has received considerable attention in recent years, studies on problems of regression are rare. This paper provides a systematic review of the most commonly used selection strategies for active learning within the context of linear regression. The recently developed Exploration Guided Active Learning (EGAL) algorithm, previously deployed within a classification context, is explored as a selection strategy for regression problems. Active learning is demonstrated to significantly improve the learning rate of linear regression models. Experimental results show that a purely diversity-based approach to
Active Learning For Text Classification, Rong Hu
Active Learning For Text Classification, Rong Hu
Doctoral
Text classification approaches are used extensively to solve real-world challenges. The success or failure of text classification systems hangs on the datasets used to train them, without a good dataset it is impossible to build a quality system. This thesis examines the applicability of active learning in text classification for the rapid and economical creation of labelled training data. Four main contributions are made in this thesis. First, we present two novel selection strategies to choose the most informative examples for manually labelling. One is an approach using an advanced aggregated confidence measurement instead of the direct output of classifiers …