InterJournal Complex Systems, 27
Status: Accepted
Manuscript Number: [27]
Submission Date: 963011
Next-generation computing and computers: data-driven, medium-based problem solving
Author(s): Val Bykovsky

Subject(s): CX.07

Category: Brief Article


State-of-the-art computer architectures and computing technologies are critically reviewed with focus on fundamental principles of computing, their limitations, and implementations. The role of pre-existing input-output pairs (IOP) or "tests" in a program building is analyzed. The emphasis is on how the pre-existing data, in fact, an initial expertise on the problem, is implicitly used to build a program which is to be viewed as a data-based estimator rather than an ab initio perfect calculator. That means that a notion of a computer that "solves" problems seems to be quite limited. We discuss, then, problem solving techniques, based on data-driven, medium-based estimation}. They provide a solid basis for robust, reliable solutions. A dynamic physical medium} and an IOP database with interpolation capability become a core of the problem solver. The IOP data is first programmed into the physical medium which then is used to estimate solutions for new inputs. Various media are briefly discussed with focus on their rich dynamics and its control. The medium-wide wavefronts ("signals") and interference patterns ("data as relationships") are considered as a basis for the next-generation media-based computing technology. Their short-term implementation using novel, wavefront-oriented silicon array or cellular VLSI architectures, including the ones with an optical pattern input-output functionality is proposed. The long-term implementation based on pre-pumped physical media is also discussed. The role of software and its evolution are briefly discussed in physical computing context.

Retrieve Manuscript
Submit referee report/comment

Public Comments: