SoQal: selective oracle questioning for consistency based active learning of cardiac signals
Clinical settings are often characterized by abundant unlabelled data and limited labelled data. This is typically driven by the high burden placed on oracles (e.g., physicians) to provide annotations. One way to mitigate this burden is via active learning (AL) which involves the (a) acquisition and...
| Main Authors: | , , |
|---|---|
| Formato: | Conference item |
| Idioma: | English |
| Publicado em: |
Journal of Machine Learning Research
2022
|
| _version_ | 1826314644752433152 |
|---|---|
| author | Kiyasseh, D Zhu, T Clifton, D |
| author_facet | Kiyasseh, D Zhu, T Clifton, D |
| author_sort | Kiyasseh, D |
| collection | OXFORD |
| description | Clinical settings are often characterized by abundant unlabelled data and limited labelled data. This is typically driven by the high burden placed on oracles (e.g., physicians) to provide annotations. One way to mitigate this burden is via active learning (AL) which involves the (a) acquisition and (b) annotation of informative unlabelled instances. Whereas previous work addresses either one of these elements independently, we propose an AL framework that addresses both. For acquisition, we propose Bayesian Active Learning by Consistency (BALC), a sub-framework which perturbs both instances and network parameters and quantifies changes in the network output probability distribution. For annotation, we propose SoQal, a sub-framework that dynamically determines whether, for each acquired unlabelled instance, to request a label from an oracle or to pseudo-label it instead. We show that BALC can outperform start-of-the-art acquisition functions such as BALD, and SoQal outperforms baseline methods even in the presence of a noisy oracle. |
| first_indexed | 2024-12-09T03:10:25Z |
| format | Conference item |
| id | oxford-uuid:cc26bf2c-a6ab-4e4e-b049-9c1893e6a610 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:10:25Z |
| publishDate | 2022 |
| publisher | Journal of Machine Learning Research |
| record_format | dspace |
| spelling | oxford-uuid:cc26bf2c-a6ab-4e4e-b049-9c1893e6a6102024-09-26T15:40:25ZSoQal: selective oracle questioning for consistency based active learning of cardiac signalsConference itemhttp://purl.org/coar/resource_type/c_5794uuid:cc26bf2c-a6ab-4e4e-b049-9c1893e6a610EnglishSymplectic ElementsJournal of Machine Learning Research2022Kiyasseh, DZhu, TClifton, DClinical settings are often characterized by abundant unlabelled data and limited labelled data. This is typically driven by the high burden placed on oracles (e.g., physicians) to provide annotations. One way to mitigate this burden is via active learning (AL) which involves the (a) acquisition and (b) annotation of informative unlabelled instances. Whereas previous work addresses either one of these elements independently, we propose an AL framework that addresses both. For acquisition, we propose Bayesian Active Learning by Consistency (BALC), a sub-framework which perturbs both instances and network parameters and quantifies changes in the network output probability distribution. For annotation, we propose SoQal, a sub-framework that dynamically determines whether, for each acquired unlabelled instance, to request a label from an oracle or to pseudo-label it instead. We show that BALC can outperform start-of-the-art acquisition functions such as BALD, and SoQal outperforms baseline methods even in the presence of a noisy oracle. |
| spellingShingle | Kiyasseh, D Zhu, T Clifton, D SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title | SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title_full | SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title_fullStr | SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title_full_unstemmed | SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title_short | SoQal: selective oracle questioning for consistency based active learning of cardiac signals |
| title_sort | soqal selective oracle questioning for consistency based active learning of cardiac signals |
| work_keys_str_mv | AT kiyassehd soqalselectiveoraclequestioningforconsistencybasedactivelearningofcardiacsignals AT zhut soqalselectiveoraclequestioningforconsistencybasedactivelearningofcardiacsignals AT cliftond soqalselectiveoraclequestioningforconsistencybasedactivelearningofcardiacsignals |