Welcome to my webpage for GPRAM

General Purpose Representation and Association Machine

What is GPRAM?

• You just need to accept the fact-the brain needs to represent the outside world in its inside and makes associations among its representations for solving general purpose problems, then GPRAM is a general purpose machine that we strip out all other functionalities of the brain.
• It will be a testing platform to develop theories, methodologies, and sharp our tools for constructing general purpose machines.

How is related to communication/IT theory?

• Ask many life scientists, “What could be structure, repeat, random permutation, nonlinear interaction, then merge?” almost all of them answered “neural structures”. But it is also the best coding structure known as Forney Normal Graph. Is error control one of key functions in neural networks? Without knowing what brain is doing, it is very hard to prove or disprove this statement, even though many remarkable similarities.
• Telecom engineers and IT scientists are now working on how to extend best known information processing algorithms to understand biological systems.

What are differences between computer and GPRAM?

• There are two ways to explore the world: (a) for a given setting and a given problem, find a precise solution (machine) to solve the problem. (b) Build general purpose machines that can vary themselves and let themselves discover the optimal setting. Our science and engineering principle follow approach (a). In these activities mathematics (often called optimization) is a very important and useful tool. GPRAM follows approach (b).
• Computer will help us to find best solutions for simple settings (like 7776 ideal templates for a simple visual task in my paper [J31]). GPRAM will discover “good” solutions for many tasks and settings (a “good” super-template for many tasks).
• A computer is measured largely by its speed and accuracy. A GPRAM should be measured by its variation.
• It has been increasing difficult to make a computer more flexible (in term of software). Equally, it will be increasing difficult to make a GPRAM more precise.

Can we build a GPRAM?

• Yes! However, it is beyond my single-person capability. Hopefully, in near future we can get resource to make it possible.