I am not sure how to characterize the math involved as it becomes a permutation of everything that I learn. I see that application of finite element analysis with Haar-like methods, vector spaces, manifolds, non-linear algebra, and every possible transform that can be associated. It is all math, but when I say that a point or section of an image is a vector in x,y,r,g,b,t,δi, and numerous other properties, it constitutes a sub space section in a new way that applies to a non-uniform manifold or has different solution sets than a simple linear solution. It may have 0, 1, infinite, or simply many solutions. The resolution of the process is toward its utility for identification, transfer and use.
In this case I am taking a video and audio stream to identify the underlying character as a solution like vector graphics and models. I can take a video of some 64 million bytes and compress it to less than a megabyte while losing no useful information. Like DNA, the information is a method which acts within the framework of universal laws of physics. I would say that DNA is a bit more imprecise in its solution sets, but still serves in its context. Transverse relativity, mitochondrial OS, information, AI, and everything else is a single complex set of relationships which are all the same in the final analysis. Solving the method of solving , leads to all the solutions.
I can perform a 2D Fourier on an image and it bears no resemblance to the original visually. I do not have the hardware in my cranium to view a Fourier set and recognize its transform product. I can however implement methods which allow me to observe and recognize the path which needs to be taken in those data sets.
Unlike the images which are compressed using lossy methods (like the ones on this blog), the key information is retained at far lower data rates than would be needed for bit-wise expression. As the amount of data that can be maintained by an individual increases exponentially, and the methods of analysis improve, they reach a point where I need not transfer data at all. I have experimented with this for my own networks and I can achieve secure communication of extremely complex information with bandwidths that would fit in 110 baud data streams without hampering the utility. It also has implications for space travel as well as gravity control. When I look at solutions that have been devised for critical functions it sometimes looks like a person with a bag on his head trying to navigate a maze. Of course that is only when I am feeling chipper, and other times I look at the universe through a bag. I am certain that information can be implement as neural structure without experimental training of the human neural net. Strangely, some resolution matrices are discovered that I only suspected and it disturbs me a bit as the implications are that something is horribly amiss in the structure of this planet's "operating system". Hopefully I will get this all HTML5 converted and anybody can use the intellectual microscope to see inside the devious structure that poses as kindness. Strange days and visions of even more bizarre futures come more often.