当前位置:网站首页>Neural structure search and an effective multi-objective evolutionary framework (CS CV)

Neural structure search and an effective multi-objective evolutionary framework (CS CV)

2020-12-07 19:23:52 Ling Qian

Deep learning has been very successful in solving many complex tasks , Such as image classification and segmentation , Speech recognition and machine translation . However , Because of the large searching space of the super parameter 、 Long training time and lack of technical guidelines for the selection of parameters , It is very difficult and time-consuming to manually design neural networks for specific problems . Besides , Most networks are highly complex , Task specific and over parameterized . lately , Multi objective neural structure search (NAS) Methods are proposed to automate accurate and efficient structural design . However , They only optimize the macro or micro structure of an architecture that requires manual definition of no hyper parameters , Instead of using the information generated in the optimization process to improve search efficiency . In this work , We have put forward EMONAS, An efficient multi-objective neural structure search framework , For automatic design of neural structures , At the same time, optimize the accuracy and scale of the network .EMONAS It consists of a search space considering both the macro and micro structure of the architecture and an agent assisted evolutionary based multi-objective algorithm , The algorithm uses random forest agents and guided selection probability to search for the best hyperparameters effectively .EMONAS adopt MICCAI ACDC The challenge assesses the three-dimensional heart segmentation task , This is the diagnosis of the disease 、 Risk assessment and treatment decisions are crucial .EMONAS The architecture found is ahead of the challenge in all metrics 10 name , Perform better or comparable to other methods , At the same time, it reduces 50% The search time above , And the number of parameters is quite small .

Original title :Neural Architecture Search with an Efficient Multiobjective Evolutionary Framework

original text :Deep learning methods have become very successful at solving many complex tasks such as image classification and segmentation, speech recognition and machine translation. Nevertheless, manually designing a neural network for a specific problem is very difficult and time-consuming due to the massive hyperparameter search space, long training times, and lack of technical guidelines for the hyperparameter selection. Moreover, most networks are highly complex, task specific and over-parametrized. Recently, multiobjective neural architecture search (NAS) methods have been proposed to automate the design of accurate and efficient architectures. However, they only optimize either the macro- or micro-structure of the architecture requiring the unset hyperparameters to be manually defined, and do not use the information produced during the optimization process to increase the efficiency of the search. In this work, we propose EMONAS, an Efficient MultiObjective Neural Architecture Search framework for the automatic design of neural architectures while optimizing the network's accuracy and size. EMONAS is composed of a search space that considers both the macro- and micro-structure of the architecture, and a surrogate-assisted multiobjective evolutionary based algorithm that efficiently searches for the best hyperparameters using a Random Forest surrogate and guiding selection probabilities. EMONAS is evaluated on the task of 3D cardiac segmentation from the MICCAI ACDC challenge, which is crucial for disease diagnosis, risk evaluation, and therapy decision. The architecture found with EMONAS is ranked within the top 10 submissions of the challenge in all evaluation metrics, performing better or comparable to other approaches while reducing the search time by more than 50% and having considerably fewer number of parameters.

Original author :Maria Baldeon Calisto, Susana Lai-Yuen

Original address :https://arxiv.org/abs/2011.04463

Original statement , This article is authorized by the author + Community publication , Unauthorized , Shall not be reproduced .

If there is any infringement , Please contact the yunjia_community@tencent.com Delete .

版权声明
本文为[Ling Qian]所创,转载请带上原文链接,感谢
https://chowdera.com/2020/11/202011190317425199.html