CIFAR-10 | famous CIFAR-10 dataset | Machine Learning library

The reason your model is taking so long to train is because:

1. You are training a large model with many layers for 100 epochs
2. You are using a relatively low performance computer (from what I found googling it only has a 2.5 GHZ processor.)

You can make it train faster by using an a free cloud environment that has GPUs and TPUs like Google Colab (https://colab.research.google.com/), or even better a Kaggle notebook which allows you to train for longer periods of time. If you want to run it on your mac you could try making the model smaller or decreasing the number of epochs you are training for.

It should be easy to port your notebook to a Google Colab or Kaggle notebook. You will need to create a google account for Google Colab or a seperate account for Kaggle.

Hope this helped!

First, the image from the docs you show is for the AE, not the VAE. The results for the VAE look much worse:

https://pl-bolts-weights.s3.us-east-2.amazonaws.com/vae/vae-cifar10/vae_output.png

Second, the docs state "Both input and generated images are normalized versions as the training was done with such images." So when you load the data you should specify normalize=True. When you plot your data, you will need to 'unnormalize' the data as well:

Yes,figured out the error. updated code below

Several problems:

• \citep is a natbib macro. If you want to use it in biblatex, you must use the natbib option when you load biblatex.

• you shouldn't load package more then once. You MUSTN'T load them more than once with different options. An error message will explicitly tell you about the option clash for the geometry package

• the syntax \begin{filecontents*}[overwrite]{\references.bib} is wrong, references.bib should just be the filename, not a (non-existent) macro

• the note field in the wikipedia entry caused come probelems, so I moved it to another field.

You need to use below code after training the model:

The input shape of CIRFAR is (32, 32, 3) but your model's input isn't taking that shape. You can try as follows for your model input.

The parameter save_freg is too large. It needs to be save_freg = training_samples // batch_size or less. Maybe try something like this:

If the hyperspectral dataset is given to you as a large image with many channels, I suppose that the classification of each pixel should depend on the pixels around it (otherwise I would not format the data as an image, i.e. without grid structure). Given this assumption, breaking up the input picture into 1x1 parts is not a good idea as you are loosing the grid structure.

I further suppose that the order of the channels is arbitrary, which implies that convolution over the channels is probably not meaningful (which you however did not plan to do anyways).

Instead of reformatting the data the way you did, you may want to create a model that takes an image as input and also outputs an "image" containing the classifications for each pixel. I.e. if you have 10 classes and take a (145, 145, 200) image as input, your model would output a (145, 145, 10) image. In that architecture you would not have any fully-connected layers. Your output layer would also be a convolutional layer.

That however means that you will not be able to keep your current architecture. That is because the tasks for MNIST/CIFAR10 and your hyperspectral dataset are not the same. For MNIST/CIFAR10 you want to classify an image in it's entirety, while for the other dataset you want to assign a class to each pixel (while most likely also using the pixels around each pixel).

Some further ideas:

• If you want to turn the pixel classification task on the hyperspectral dataset into a classification task for an entire image, maybe you can reformulate that task as "classifying a hyperspectral image as the class of it's center (or top-left, or bottom-right, or (21th, 104th), or whatever) pixel". To obtain the data from your single hyperspectral image, for each pixel, I would shift the image such that the target pixel is at the desired location (e.g. the center). All pixels that "fall off" the border could be inserted at the other side of the image.
• If you want to stick with a pixel classification task but need more data, maybe split up the single hyperspectral image you have into many smaller images (e.g. 10x10x200). You may even want to use images of many different sizes. If you model only has convolution and pooling layers and you make sure to maintain the sizes of the image, that should work out.

You can use ConcatDataset from torch.utils.data module.

Code Snippet: