{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/vos-a-method-for-variational-oversampling-of","title":"VOS: a Method for Variational Oversampling of Imbalanced Data","arxiv_id":"1809.02596","date":"2018-09-07","proceeding":null,"authors":["Val Andrei Fajardo","David Findlay","Roshanak Houmanfar","Charu Jaiswal","Jiaxi Liang","Honglei Xie"],"abstract":"Class imbalanced datasets are common in real-world applications that range\nfrom credit card fraud detection to rare disease diagnostics. Several popular\nclassification algorithms assume that classes are approximately balanced, and\nhence build the accompanying objective function to maximize an overall accuracy\nrate. In these situations, optimizing the overall accuracy will lead to highly\nskewed predictions towards the majority class. Moreover, the negative business\nimpact resulting from false positives (positive samples incorrectly classified\nas negative) can be detrimental. Many methods have been proposed to address the\nclass imbalance problem, including methods such as over-sampling,\nunder-sampling and cost-sensitive methods. In this paper, we consider the\nover-sampling method, where the aim is to augment the original dataset with\nsynthetically created observations of the minority classes. In particular,\ninspired by the recent advances in generative modelling techniques (e.g.,\nVariational Inference and Generative Adversarial Networks), we introduce a new\noversampling technique based on variational autoencoders. Our experiments show\nthat the new method is superior in augmenting datasets for downstream\nclassification tasks when compared to traditional oversampling methods.","url_abs":"http://arxiv.org/abs/1809.02596v1","url_pdf":"http://arxiv.org/pdf/1809.02596v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"vos-a-method-for-variational-oversampling-of","repo_url":"https://github.com/HongleiXie/demo-CVAE","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"fraud-detection","task_name":"Fraud Detection"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}