Context

CIFAR-10 features collected from VGG19 model at 85% test accuracy. Consists of features from all 19 layers of 10,000 test images. Each layer has its own directory. Inside each layer directory, the features are stored per batch with filename started from 1 to (in current case) 5 with batch size of 2,000 images.

Image dataset is from version 1. Refer to this dataset to get the image, the label and the filename. Concatenate batch files in an ascending manner to get index to index relation between the image and the feature.

Extracted from notebook version 10.

Content

16 feature layer directories (conv1_1, conv1_2, ..., conv5_4) and 3 classifier layer directories (lin1, lin2, and lin3). Each layer directory consists of saved files 1 to 5 in serialized NumPy using Python pickle. The file with 2,000 batch size has shape of [2000, dim1, dim2, dim3] for feature layers or [2000, dim1] for classifier layers.

Example for feature layer: conv1_1 (2000, 64, 32, 32) conv1_2 (2000, 64, 32, 32) conv2_1 (2000, 128, 16, 16) conv2_2 (2000, 128, 16, 16) conv3_1 (2000, 256, 8, 8) conv3_2 (2000, 256, 8, 8) conv3_3 (2000, 256, 8, 8) conv3_4 (2000, 256, 8, 8) conv4_1 (2000, 512, 4, 4) conv4_2 (2000, 512, 4, 4) conv4_3 (2000, 512, 4, 4) conv4_4 (2000, 512, 4, 4) conv5_1 (2000, 512, 2, 2) conv5_2 (2000, 512, 2, 2) conv5_3 (2000, 512, 2, 2) conv5_4 (2000, 512, 2, 2)

Example for classifier layer: lin1 (2000, 4096) lin2 (2000, 4096) lin3 (2000, 10)

The CIFAR-10.1 dataset is a new test set for CIFAR-10. CIFAR-10.1 contains roughly 2,000 new test images that were sampled after multiple years of research on the original CIFAR-10 dataset. The data collection for CIFAR-10.1 was designed to minimize distribution shift relative to the original dataset. We describe the creation of CIFAR-10.1 in the paper "Do CIFAR-10 Classifiers Generalize to CIFAR-10?". The images in CIFAR-10.1 are a subset of the TinyImages dataset. There are currently two versions of the CIFAR-10.1 dataset: v4 and v6.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('cifar10_1', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

https://storage.googleapis.com/tfds-data/visualization/fig/cifar10_1-v4-1.1.0.png" alt="Visualization" width="500px">

A re-labeled version of CIFAR-10's test set with soft-labels coming from real human annotators. For every pair (image, label) in the original CIFAR-10 test set, it provides several additional labels given by real human annotators as well as the average soft-label. The training set is identical to the one of the original dataset.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('cifar10_h', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

https://storage.googleapis.com/tfds-data/visualization/fig/cifar10_h-1.0.0.png" alt="Visualization" width="500px">

This dataset is just like the CIFAR-10, except it has 100 classes containing 600 images each. There are 500 training images and 100 testing images per class. The 100 classes in the CIFAR-100 are grouped into 20 superclasses. Each image comes with a "fine" label (the class to which it belongs) and a "coarse" label (the superclass to which it belongs).

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('cifar100', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

https://storage.googleapis.com/tfds-data/visualization/fig/cifar100-3.0.2.png" alt="Visualization" width="500px">

This repository contains the Python code required to reproduce the simulation part of the paper "On the rate of convergence of image classifiers based on convolutional neural networks" from Kohler, Krzyżak, and Walter (2022) referenced below. The Python version used is Python 3.9.7. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under project number 449102119. The Cifar-10 image dataset consisting of the real images from which the classes "cars" and "ships" were used can be downloaded from the link given below. In the Techincal Report "Learning Multiple Layers of Features from Tiny Images" from Alex Krizhevsky (2009) (for a link see below) this dataset of real images is described in more detail.