{"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/evolving-deep-convolutional-neural-networks","title":"Evolving Deep Convolutional Neural Networks for Image Classification","arxiv_id":"1710.10741","date":"2017-10-30","proceeding":null,"authors":["Yanan Sun","Bing Xue","Mengjie Zhang","Gary G. Yen"],"abstract":"Evolutionary computation methods have been successfully applied to neural\nnetworks since two decades ago, while those methods cannot scale well to the\nmodern deep neural networks due to the complicated architectures and large\nquantities of connection weights. In this paper, we propose a new method using\ngenetic algorithms for evolving the architectures and connection weight\ninitialization values of a deep convolutional neural network to address image\nclassification problems. In the proposed algorithm, an efficient\nvariable-length gene encoding strategy is designed to represent the different\nbuilding blocks and the unpredictable optimal depth in convolutional neural\nnetworks. In addition, a new representation scheme is developed for effectively\ninitializing connection weights of deep convolutional neural networks, which is\nexpected to avoid networks getting stuck into local minima which is typically a\nmajor issue in the backward gradient-based optimization. Furthermore, a novel\nfitness evaluation method is proposed to speed up the heuristic search with\nsubstantially less computational resource. The proposed algorithm is examined\nand compared with 22 existing algorithms on nine widely used image\nclassification tasks, including the state-of-the-art methods. The experimental\nresults demonstrate the remarkable superiority of the proposed algorithm over\nthe state-of-the-art algorithms in terms of classification error rate and the\nnumber of parameters (weights).","url_abs":"http://arxiv.org/abs/1710.10741v3","url_pdf":"http://arxiv.org/pdf/1710.10741v3.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":"evolving-deep-convolutional-neural-networks","repo_url":"https://github.com/MagnusCaligo/EvoCNN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"heuristic-search","task_name":"Heuristic Search"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1710.10741","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}