{"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/dropmax-adaptive-variational-softmax","title":"DropMax: Adaptive Variational Softmax","arxiv_id":"1712.07834","date":"2017-12-21","proceeding":"NeurIPS 2018 12","authors":["Hae Beom Lee","Juho Lee","Saehoon Kim","Eunho Yang","Sung Ju Hwang"],"abstract":"We propose DropMax, a stochastic version of softmax classifier which at each\niteration drops non-target classes according to dropout probabilities\nadaptively decided for each instance. Specifically, we overlay binary masking\nvariables over class output probabilities, which are input-adaptively learned\nvia variational inference. This stochastic regularization has an effect of\nbuilding an ensemble classifier out of exponentially many classifiers with\ndifferent decision boundaries. Moreover, the learning of dropout rates for\nnon-target classes on each instance allows the classifier to focus more on\nclassification against the most confusing classes. We validate our model on\nmultiple public datasets for classification, on which it obtains significantly\nimproved accuracy over the regular softmax classifier and other baselines.\nFurther analysis of the learned dropout probabilities shows that our model\nindeed selects confusing classes more often when it performs classification.","url_abs":"http://arxiv.org/abs/1712.07834v5","url_pdf":"http://arxiv.org/pdf/1712.07834v5.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":"dropmax-adaptive-variational-softmax","repo_url":"https://github.com/haebeom-lee/dropmax","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"dropmax-adaptive-variational-softmax","repo_url":"https://github.com/OpenXAIProject/UncertintyAttention_DropMax","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"dropmax-adaptive-variational-softmax","repo_url":"https://github.com/yamad07/DropMax","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"dropmax-adaptive-variational-softmax","repo_url":"https://github.com/zsef123/DropMax-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}