{"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/regularized-evolution-for-image-classifier","title":"Regularized Evolution for Image Classifier Architecture Search","arxiv_id":"1802.01548","date":"2018-02-05","proceeding":null,"authors":["Esteban Real","Alok Aggarwal","Yanping Huang","Quoc V. Le"],"abstract":"The effort devoted to hand-crafting neural network image classifiers has\nmotivated the use of architecture search to discover them automatically.\nAlthough evolutionary algorithms have been repeatedly applied to neural network\ntopologies, the image classifiers thus discovered have remained inferior to\nhuman-crafted ones. Here, we evolve an image classifier---AmoebaNet-A---that\nsurpasses hand-designs for the first time. To do this, we modify the tournament\nselection evolutionary algorithm by introducing an age property to favor the\nyounger genotypes. Matching size, AmoebaNet-A has comparable accuracy to\ncurrent state-of-the-art ImageNet models discovered with more complex\narchitecture-search methods. Scaled to larger size, AmoebaNet-A sets a new\nstate-of-the-art 83.9% / 96.6% top-5 ImageNet accuracy. In a controlled\ncomparison against a well known reinforcement learning algorithm, we give\nevidence that evolution can obtain results faster with the same hardware,\nespecially at the earlier stages of the search. This is relevant when fewer\ncompute resources are available. Evolution is, thus, a simple method to\neffectively discover high-quality architectures.","url_abs":"http://arxiv.org/abs/1802.01548v7","url_pdf":"http://arxiv.org/pdf/1802.01548v7.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":"regularized-evolution-for-image-classifier","repo_url":"https://github.com/tally0818/NASNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"regularized-evolution-for-image-classifier","repo_url":"https://github.com/xuanhungho/nsga","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"regularized-evolution-for-image-classifier","repo_url":"https://github.com/DataCanvasIO/Hypernets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"regularized-evolution-for-image-classifier","repo_url":"https://github.com/anonDEHB/DEHB","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"regularized-evolution-for-image-classifier","repo_url":"https://github.com/tensorflow/tpu/tree/master/models/official/amoeba_net","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"evolutionary-algorithms","task_name":"Evolutionary Algorithms"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"architecture-search","task_name":"Neural Architecture Search"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"}],"methods":[{"method_slug":"aging-evolution","method_name":"Aging Evolution"},{"method_slug":"amoebanet","method_name":"AmoebaNet"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"rmsprop","method_name":"RMSProp"},{"method_slug":"sgd-with-momentum","method_name":"SGD with Momentum"},{"method_slug":"scheduleddroppath","method_name":"ScheduledDropPath"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"spatially-separable-convolution","method_name":"Spatially Separable Convolution"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[{"slug":"aging-evolution","name":"Aging Evolution","full_name":"Aging Evolution"}],"results":[{"leaderboard":"/sota/image-classification-on-imagenet","task":"Image Classification","dataset":"ImageNet","model":"AmoebaNet-A","rank_in_archive_order":382,"of":1060,"metrics":{"GFLOPs":"208","Number of params":"469M","Top 1 Accuracy":"83.9%"},"uses_additional_data":false},{"leaderboard":"/sota/architecture-search-on-cifar-10-image","task":"Neural Architecture Search","dataset":"CIFAR-10 Image Classification","model":"AmoebaNet-B + c/o","rank_in_archive_order":8,"of":19,"metrics":{"Params":"34.9M","Percentage error":"2.13"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-nas-bench-201","task":"Neural Architecture Search","dataset":"NAS-Bench-201, ImageNet-16-120","model":"REA","rank_in_archive_order":25,"of":49,"metrics":{"Accuracy (Test)":"45.54","Search time (s)":"12000"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-nats-bench-1","task":"Neural Architecture Search","dataset":"NATS-Bench Topology, CIFAR-10","model":"RE (Real et al., 2019)","rank_in_archive_order":2,"of":11,"metrics":{"Test Accuracy":"94.13"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-nats-bench-2","task":"Neural Architecture Search","dataset":"NATS-Bench Topology, CIFAR-100","model":"RE (Real et al., 2019)","rank_in_archive_order":4,"of":11,"metrics":{"Test Accuracy":"71.40"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-nats-bench","task":"Neural Architecture Search","dataset":"NATS-Bench Topology, ImageNet16-120","model":"RE (Real et al., 2019)","rank_in_archive_order":5,"of":11,"metrics":{"Test Accuracy":"44.76"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.01548","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}