https://github.com/jakobzmrzlikar/meta-ml

A meta-ML model for predicting the accuracy of deep neural networks on certain datasets.
https://github.com/jakobzmrzlikar/meta-ml

deep-learning machine-learning meta-learning

Last synced: 9 days ago
JSON representation

A meta-ML model for predicting the accuracy of deep neural networks on certain datasets.

• Host: GitHub
• URL: https://github.com/jakobzmrzlikar/meta-ml
• Owner: jakobzmrzlikar
• Created: 2018-07-16T12:12:50.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2022-05-26T20:23:18.000Z (over 2 years ago)
• Last Synced: 2023-03-04T13:14:51.639Z (over 1 year ago)
• Topics: deep-learning, machine-learning, meta-learning
• Language: Python
• Homepage:
• Size: 4.37 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# meta-ML

## A meta-ML model for predicting the accuracy of deep neural networks on certain datasets. Formerly part of the [DataBench](https://www.databench.eu/) project.

### Dependencies:

*  [Keras](https://github.com/keras-team/keras)

*  [Tensorflow](https://www.tensorflow.org/)

*  [numpy](https://www.numpy.org/)

*  [scikit-learn](https://scikit-learn.org/stable/)

*  [psutil](https://pypi.org/project/psutil/)

### Usage

The _config_ directory contains JSON files. You must first specify your model's features and which dataset to use. Leave the _results_ section empty. After you input all the necessary parameters, simply run main.py and the _results_ section will be automatically updated. The metadataset is automatically updated as well each time a model is run on a dataset.

You can alternatively only specify certain model architecture parameters and run config_generator.py to generate new config files with different hyperparameter values.