Our company
Patents

Data Invest is involved in the development of advanced technologies, as well as their effective implementation in the business. Below there are the technologies that are patent pending:

Patent 1

The way of analyzing and the mechanism for analyzing synchronous signals dependable on time

 

Patent 2

A way of structuring and functioning of metadata in semantic networks

 

Patent 3

Manner of Data-Perspectives-Logic-Views-Templates ( DPLVT )

 

Patent 4

Way of evolution algorithms popularization with the use of coding genotype of individual neuron nets

 

Patent 5

Way of processing genetic algorithms equipped with individual neuron nets with the use of many units GPGPU

 
The patent connected with huge data set processing in the time comparable to super computers with simultaneous saving costs related both to processor price ( simultaneous use of GP and CP units ) and servers that they operate on and to energy use, too. Utility of counting techniques of this kind will let modern counting methods develop, including Artificial Intelligence algorithms because of high parallelism of counting processes. Patent rights guarantee the way adjusted to accounting units ( GPU ) while analyzing the mass inflow of data in the real time ( for example: stock quotations, currency fluctuation rate, stock rating ). Patents may become useful in nonfinancial matters ( medicine, bio – and nano – technology, neural science, seismography, etc ). Examples of areas of business usage:
  • Analyzing of huge finance data set, for example from capital markets, raw material markets, etc.
  • Fast analyses of huge data streams related to sale, currency and price fluctuation, number of various factors and parameters ( not only financial but technical in electro – engineering networks ).
  • Analyses of various dynamic models ( strictly dependant on time or other parameter ), their simulation ( also in the pattern ‘what –if “ – simultaneous counting of many alternatives ), e.g. financial models, market simulations, risk assessment, marking out some dependency between changeable in time parameters, grid protection from excessive load, transport processes’ optimization, wind – power plants’ simulations, etc ).
  • Providing other analytic systems with great accounting capacity adjusted to mass collateral assessments ( which means that it allows to make a great deal of transactions in the same time ).
  • Signal analyses, e.g. medical ( coming from diagnostic equipment ) and their processing in order to find dependency and visualization too.
  A patent lets data bases be built. These data bases contain additional information ( metadata ) connected with stuff intended for mobile phones and other mobile equipment. Stuff mentioned here: e.g. rings, wallpapers, games, programs. Whereas the metadata are: descriptions, target groups, sale efficiency, association to other problems. The core of patent rights is the way of storing ( as a semantic net ) the additional data ( metadata ), allowing for intelligent analyzing with the use of Artificial Intelligence methods ( e.g. classification, heuristic deduction, genetic algorithms, decisive trees ). To examine and predict consumer trends, an in-coming offered way of structuring metadata will be used.
Business usage:
  • Fast processing of metadata related to content ( in the real time ) in order to adjust to consumer needs as fast as possible, e.g. joining various contents into lower price packages.
  • The use of parallel units allows of multidimensional analysis which are better than ”classical” BI with regards to speed, automatic “understanding” of the meaning of analyzed content ( the content is semantic, not only numeral ) and pursuit of optimization processes in an inaccessible way with explicit approach ( which is in classical BI ).
  • Possible cooperation with mobile phone software in order to examine consumer behavior and reach target consumer groups and, what’s more possible, the process of building of appropriate consumer archeotypes.
  The core of the invention is the manner of building program applications with the use of additional data layer which allows for preventing intact structure and content of basic data layer while changing the logic of software action. In a standard model of creating programming applications ( Model-View-Controller) it is assumed that a model layer also includes access to data and enables operating on them. Hence it can be treated as data connection, operations carried on them and their structure with application logics. The problem appears when making changes in logics that force modification of data layer ( usually on the level of data base structure ). It reflects in significant costs related to application development, particularly if software is complicated and present on the market for a longer time. The aim of building programming application allowing for essential extension of profitability periods of making their multiple modifications, particularly beyond achievable time of product life. The aim was realized through the elaboration of the manner of building programming applications in the view of multilayer model comprising data, perspectives, logic, views and templates.

Business utility:
  • Making changes in software without the necessity to modify the structure and content of source data which results in savings in project.
  • Enlarging flexibility of building IT solutions, which allows for future solid reconstruction of software logic.
  • Possibility to build separate data layers and perspectives layers by independent programming teams, which influences cutting the time of creating an application and simultaneously the costs of its creation and reconstruction.
Genetic algorithms ( in other words evolution ) are one of the most advanced method of Artificial Intelligence ( AI ). They are used to solve ambiguous problems ( usually complex and insufficiently determined ). The principle of the patent is the use of genetic code and natural selection rules ( their virtual counterparts are “sewed” in particular genetic algorithms and the environment of their action ). Algorithms, that solve some part of a problem, are associated among themselves ( genotype population ), the weaker are removed ( natural selection ). Next “generations” of algorithms cope with a problem better and better and bring into effect their whole solution or come very near to their solution ( when optimal solution isn’t achievable ). Patent “reinforces” genetic algorithms with another object of AI – neuron nets, which are also coded with the use of appropriate part of virtual genotype. Genotype, characteristic for genetic algorithms, is then used for developing and coding neuron nets ( here in another generations partially trained nets are multiplied, not algorithms in explicite form ).

Business utility:
  • Solving complex calculating tasks on which particular algorithms are not created, i.e. investment portfolio optimization, logistic problems, risk management, complex optimizing processes, searching information, facts association in expert systems etc.
  • Social consumer behaviors’ simulations, fashion phenomena and their passing, etc., what’s particularly important when building competition dominance on the market dominated by consumers’ preferences and opinions.
  • Scanning huge data and metadata sets and finding on their bases existing dependence, usually invisible in any simplified set.
  This patent is a developing of the previous one with the use of specialist multi-spinal processors ( GPGPU ), originally used for graphic operations. The data processing manner with the use of these processes much differs from classical methods. Hence the method of genetic algorithms software equipped with individual neuron nets using multi – spinal processors has been worked out. It allows not only for high acceleration of counting but it enables rational savings connected with lower cost of purchase of calculating power and its higher efficiency for energy consumption. Using the environment for parallel/simultaneous calculating, that GPGPU items are adjusted to, is very significant. Using GPGPU gives radical advantage because i.e. neuron calculating use mass parallel but in most cases simple operations. So the amount of simpler ( and cheaper ) calculating items is more important than calculating power of single spinals as it happens in case of classical CPU which are used in relatively expensive super computers and clusters.

Business utility:
  • As for patent entitled “Way of evolution algorithms popularization with the use of coding genotype of individual neuron nets”, however with a difference that calculating power of parallel items is used.