Training and test data, plus saved models for the paper "Top-down inference in an early visual cortex inspired hierarchical Variational Autoencoder" submitted to NeurIPS 2022

Each .pkl file contains a training or test dataset in the form of a Python dictionary (generated with Python 3.8.5) with the following fields:

'train_images': 640,000 float32 images used for model training. 20px images contain 400 pixel intensities, 40px images contain 1600 pixel intensities each.

'train_labels': float32 labels for each image in 'train_images'. All natural images are labeled with 0.0. Texture images are labeled with 0.0, 1,0, 2.0, 3.0, or 4.0, according to their texture family.

'test_images': 64,000 float32 images used for model testing. 20px images contain 400 pixel intensities, 40px images contain 1600 pixel intensities each.

'test_labels': float32 labels for each image in 'test_images'. All natural images are labeled with 0.0. Texture images are labeled with 0.0, 1,0, 2.0, 3.0, or 4.0, according to their texture family.

Each .zip file contains a saved model. Details on these are coming soon.

For more details, see the paper "Top-down inference in an early visual cortex inspired hierarchical Variational Autoencoder" submitted to NeurIPS 2022 (preprint).