mat | R demo package illustrating BLAS and LAPACK calls

Apparently V7.3 is not the only incompatible mat file version. As stated in the RDocuments,

readMat: Reads a MAT file structure from a connection or a file Description

Reads a MAT file structure from a connection or a file. Both the MAT version 4 and MAT version 5 file formats are supported.

And long story short, version 4 and version 5 cannot save such large dataset in one file. I think at least 2 solutions are straight forward:

1. Exchange data in a different file format, e.g. HDF5 or SQlite. Such files are well supported both in R and matlab and do not have the compatibility issue.

2. Save mat file in matlab in version 4 with the '-v4' switch, but there is an upper size limit in version 4, so you'll likely need to split you data across multiple files.

You do not need loops here. Just use the whole matrix in your call to x>5

The tricky part is summing the diagonals. I've questioned multiple times in the past about the fastest approach to accomplish this.

One solution is to use bincount:

Use replace

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.

Contrary to common belief Haskell is quite friendly with respect to problems like that. The real issue is that the array library that comes with GHC is total garbage. Another big problem is that everyone is taught in Haskell to use lists where arrays should be used instead, which is usually one of the major sources of slow code and memory bloated programs. So, it is not surprising that GC takes a long time, it is because there is way too much stuff being allocation. Here is a run on the supplied input for the solution provided below:

TL;DR
It have to do with the StandardScaler, change it to a simple normalisation.
e.g.

In the past, many developers would add autocomplete="off" to their form fields to prevent the browser from performing any kind of autocomplete functionality. While Chrome will still respect this tag for autocomplete data, it will not respect it for autofill data.

One workaround is to put an unknown value in the autocomplete,

.

When testing this it worked for me most of the time, but for some reason didn't work anymore afterwards.

My advise is not to fight against it and use it's potential by properly using the autocomplete attribute as explained here.

Those styles are rewritten. Use !important:

styles.scss: