No products are associated with this question.
Probably not much. You can reduce the number of directories in your matlab path to reduce some of the startup time. You can pcode your files to reduce execution time a little (probably not much unless your code is huge.)
You can make sure your computer has lots of memory, and have something that reads all of your files ahead of time so that they are in disc cache. This is done automatically on MS Windows systems in quite new versions of MATLAB.
If you are accessing hard-disk, you can move everything onto solid state storage.
The profiler reveals, where the most time is spent. I suggest to find this bottleneck and post the code here.
In a complex GUI with a lot of elements, the rendering cna consume a lot of time. So add a drawnow inside the code, otherwise the profiler cannot consider the correpsonding time, because the rendering is started, when the control returns to the command line and therefore after the profiler has finished the measurements.