René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Issue with Macro Average F1 Score in modAL for Multi-Class Classification","articleBody":"Hello,\r\nI am currently using the modAL library to develop active learning strategies. My project involves multi-class classification implemented with PyTorch, utilizing DinoV2 as the backbone model. I have encountered an issue with the macro average F1 score in my active learning loop.\r\n\r\nWhen I used the macro average F1 score, the results were unsatisfactory. The active learning loop remained stuck at one F1 score value during each iteration. However, when I switched to the micro average F1 score, the results improved significantly.\r\n\r\nI am confident in the quality of my dataset and the factors influencing it, as it performs well with other strategies using multi-class classification and the macro F1 score. Therefore, I suspect there might be an issue related to modAL's handling of the macro average F1 score.\r\n\r\nI would like to know if there are any known inconsistencies or issues with using macro averaging in modAL when working on multi-class classification.\r\n\r\nThank you for your guidance in advance.","author":{"url":"https://github.com/shadikhamsehh","@type":"Person","name":"shadikhamsehh"},"datePublished":"2024-07-20T17:07:09.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/188/modAL/issues/188"}