UMTRA on Mini-Imagenet

This is the code for applying unsupervised meta-learning UMTRA algorithm on Mini-Imagenet dataset. We build this on top of the code for Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks (Finn et al., ICML 2017).

Dependencies

This code requires the following:

• python 2.* or python 3.*
• TensorFlow v1.0+
• Tensorflow-Hub (pip install --upgrade tensorflow-hub==0.7.0)

Data

For the MiniImagenet data, see the usage instructions in data/miniImagenet/proc_images.py. For applying UMTRA on other datasets use this link.

Usage

To run the code, see the usage instructions at the top of main.py. Use train=True for training and then after train is finished set train=False and test_set=True in order to use test set. Notice that this code uses UMTRA algorithm for train.

Contact

To ask questions or report issues, please open an issue on the issues tracker.