How to build:

The project has TBB and OpenMP as dependencies, but I believe any Linux installation
should have these available by default.

Build the project using CMake. Simply running `cmake -DCMAKE_BUILD_TYPE=Release -Bbuild .` followed by `make`
in the project's root directory should suffice to produce the binaries within build/ .

How to run:

Make sure to decompress mnist_train.tar.gz into working directory you will run `train` in.

Then, just run said program. Ctrl + C to quit, stop training and produce "parameters.dat"
at any time. You can also send SIGUSR1 to the process if you know its PID. This will
produce parameters.dat, but the program will keep on running.

To generate samples from the produced parameters.dat, simply run reconstruct. It will produce
a file called output.bmp .

BEWARE!!!! Both programs will overwrite their corresponding output files when run again, so
make sure to back up any important results before doing so!