Back in 1811 there was a mathematical technique proposed, it now known to us as Joseph Fourier’s Fourier Technique, a technique which eventually translated itself into one of the most basic ways we now encode and de-code images, music, and many other kinds of media – what’s being proposed by MIT now is that there’s a much more efficient way to go about all of it. The Fourier Technique took a complex signal, broke it down into several components, transmitted or processed them each separately, then re-combined them into what was basically there in the first place. In 1965 this technique was translated to computers, and sweet packaging did occur.
Now we’re taking multiple pieces of a file, telling the computer to see each similar piece as one piece of information, essentially, and sending. As publisher Devin Coldewey so elegantly puts it: instead of sending a music file over to your friend instrument by instrument, you just stack the sounds on top of one another, this resulting in a single file much more complicated than any of the individuals. This can all be simplified, says MIT researchers working to improve this “discrete” Fourier transform.
What they’ve found, or what they’re showing in the paper about to be published on the matter, is that the level of simplification that can be applied to this process can be rather drastic. For example they have an 8×8 block of values (64) in which 57 can be discarded – this happening without visibly affecting the file at hand, the file in this case being an image. What this could mean for our future is some gigantic things, ironically, especially in the fields of mobile computing and home computing – entertainment to the max. Have a peek at the paper as it’s titled now: Nearly Optimal Sparse Fourier Transform written up by Haitham Hassanieh, Piotr Indyk, Dina Katabi, and Eric Price.