T-Bit: a unit of information 1000x of times more efficient_post_30

Company

Company

T-Bit: a unit of information 1000x of times more efficient

The modern scientific language in general, and the language of mathematics in particular are not suitable for describing the processes (computability) of the brain. The brain does not use integrals for calculations, nor does it use symbols (letters, numbers) that we are accustomed to. It applies some other elements, and the laws by which these elements interact with each other are different from mathematical and linguistic laws. The brain uses another kind of computability, a “non-mathematical” or even “non-scientific” one.

01

04

TAPe-based technologies will operate with representations rather than arrays of structurally incoherent numbers (zeros and ones) that are used now. A representation, or T-bit as we have called them, is a description involving a subset of the most informative connected elements of information.

07

Nowadays, computer vision technologies developed with TAPe are already more efficient than existing developments in this field both in terms of accuracy and resource capacity, as well as in terms of versatility.

02

This specific computability of the brain could be called the very Language of Thought that Jerry Fodor was talking about. Nevertheless, Fodor just came up with a term that we think is appropriate, yet he did not develop any theory. It is the Theory of Active Perception, TAPe, that describes at least part of the laws and methods of brain computability. TAPe is distinct from both traditional mathematics and linguistics, but it “combines” their methods and laws. TAPe simulates the workings of innate perceptual mechanisms of a human.

05

We must not hope for a technology to be developed that will be as efficient as natural intelligence. However, we believe that TAPe is by orders of magnitude closer to that than any other current technology, including AI. TAPe simulates how the human brain perceives information, and its methods can be used to develop products and technologies that are superior in efficiency to the ones that exist.

08

However, TAPe is applicable far beyond computer vision technologies. With TAPe, we can create new principles of construction and architecture of neural systems and computer processors, ALUs, IT devices in general, new data centers, methods of working with videos that allow solving any CV tasks without additional transformations, etc.

06

Unlike AI, in new types of algorithms, thanks to TAPe, the same primary data will allow us to solve problems of various categories from completely different industries. TAPe's mathematical framework is the same for any task. T-bits do not differ in information types, i.e. the same T-bit can describe absolutely different types of information including images, texts, videos or sounds. This is why we call the technology superior: when we apply TAPe methods to all kinds of tasks, the number of computational operations is reduced by orders of magnitude.

03

TAPe represents a new way of information processing. It is based on a model that replaces the binary number system of 0 and 1. This allows a unit of information (bit) to convey much more meaningful information, making it possible to create new technologies based on TAPe methods to process any kind of information.

09

Moreover, we are convinced that TAPe allows us to make a qualitative leap in setting and solving practical tasks. We will be able not only to deal with current tasks more efficiently than with current technologies, but also to set completely different tasks that cannot be solved with the help of today's technologies. This includes building algorithms and/or methods on the basis of TAPe, which will make it possible to reach the efficiency of natural intelligence, virtually using its own methods.