https://github.com/zh-plus/evolutionary_neural_network-stop

Using evolutionary algorithm to search for a great neural network architecture.
https://github.com/zh-plus/evolutionary_neural_network-stop

Last synced: 3 months ago
JSON representation

Using evolutionary algorithm to search for a great neural network architecture.

• Host: GitHub
• URL: https://github.com/zh-plus/evolutionary_neural_network-stop
• Owner: zh-plus
• License: mit
• Created: 2018-10-14T05:09:23.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-12-10T11:05:00.000Z (over 6 years ago)
• Last Synced: 2025-02-28T08:29:46.999Z (4 months ago)
• Language: Python
• Homepage:
• Size: 61.5 KB
• Stars: 5
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        


##### Current Situation

​Basically, we'll use MNIST as our test bench now.

​The preliminary goal we want to achieve is evolving an neural network that can rival the [model](https://github.com/zh-plus/NN_example/blob/master/CNN/mnist_highACC.py) in my NN_example repository, i.e. approximately 99.6 test accuracy in MNIST data set.

In the future, we will also adapt it into more data set, namely CIFAR-10, CIFAR-100 and ImageNet.

##### References

- [Large-Scale Evolution of Image Classifiers](https://arxiv.org/pdf/1703.01041.pdf)

- [Regularized Evolution for Image Classifier Architecture Search](https://arxiv.org/pdf/1802.01548.pdf)

- [Learning Transferable Architectures for Scalable Image Recognition](https://arxiv.org/pdf/1707.07012.pdf)