{"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/automatic-configuration-of-deep-neural","title":"Automatic Configuration of Deep Neural Networks with EGO","arxiv_id":"1810.05526","date":"2018-10-10","proceeding":null,"authors":["Bas van Stein","Hao Wang","Thomas Bäck"],"abstract":"Designing the architecture for an artificial neural network is a cumbersome\ntask because of the numerous parameters to configure, including activation\nfunctions, layer types, and hyper-parameters. With the large number of\nparameters for most networks nowadays, it is intractable to find a good\nconfiguration for a given task by hand. In this paper an Efficient Global\nOptimization (EGO) algorithm is adapted to automatically optimize and configure\nconvolutional neural network architectures. A configurable neural network\narchitecture based solely on convolutional layers is proposed for the\noptimization. Without using any knowledge on the target problem and not using\nany data augmentation techniques, it is shown that on several image\nclassification tasks this approach is able to find competitive network\narchitectures in terms of prediction accuracy, compared to the best\nhand-crafted ones in literature. In addition, a very small training budget (200\nevaluations and 10 epochs in training) is spent on each optimized architectures\nin contrast to the usual long training time of hand-crafted networks. Moreover,\ninstead of the standard sequential evaluation in EGO, several candidate\narchitectures are proposed and evaluated in parallel, which saves the execution\noverheads significantly and leads to an efficient automation for deep neural\nnetwork design.","url_abs":"http://arxiv.org/abs/1810.05526v1","url_pdf":"http://arxiv.org/pdf/1810.05526v1.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":"automatic-configuration-of-deep-neural","repo_url":"https://github.com/wangronin/MIP-EGO","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"global-optimization","task_name":"global-optimization"},{"task_slug":"image-classification","task_name":"image-classification"}],"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}