DESC:Technical field
The present invention pertains to the artificial digital neurons. More particularly, to the creation of the memory patterns using defined length blocks of data.
Background of the invention
In the biological neuron system, the neuron are the basic unit which are responsible for carrying and storing the information pertaining to any action, reaction, stimuli or the response of the body. With every perception from the sensory organs, a specific set of neurons are fired or triggered which creates a particular string of sequence that can be referred to as a pattern. When the same perception for the same thing is received the next time, the same sets of neurons are triggered with same pattern which is stored as the memory. The pattern is thus coded inside the brain in the form of a memory wherein the pattern itself codes for the information perceived.
Artificial neural networks are powerful information processors conceptually based on the field of neurobiology as is well known in the art. Such networks strive to simulate biological neural network functions in computer architectures. In other words, an artificial neural network must correlate a wide variety of inputs to produce an output.
The invention such as mentioned in US patent US5673367A, the neural network is used to control the motion in robotics wherein the control is imparted to the robot through training sets that triggers its sensor for the specific motion. However this system is time consuming and limits the motion of the robotics as it is capable of movements only imparted in the training sets.
The present invention provides the real-time learning and memorizing of the information such as pertaining to any type of motion like walking, running etc or actions like thinking, dreaming etc. This eliminates the need to impart the training sets in case of the AI or robotics and widens the scope of information or the data perceived and interpreted.
Summary of the invention
The present invention pertains to the digital neurons wherein the digital neurons are used to create memory patterns using a coding system wherein any unit of metric system can be used for coding, for example HEX coding system, binary system, octal system, base 10 system, base 2 system, cryptographic hash and like.
The present invention mimics the neural functioning of the biological system to store and memorize the information perceived by the sensory systems wherein such information is stored in the form of Unit of storage for example HEX codes in variable or fixed data storage length.
In the Hex coding system, HEX value is an 8 digit code (4 hex code till 255) starting from 00000000 to ffffffff. Each hex code (with decimal value from 0 to 255) represented by first two digits such that first two digits are is one neuron. So the example mentioned above, we could have 4,228,250,625 unique neurons (more than 4 billion neurons) and their combinations. Thus any information perceived can be HEX coded without any chance of repetition of the code. The HEX coding can change to any other unit and total digit code can change from 8 to 16 or even more than 16 and less than 8. This creates a scenario as in actual brain where data is stored. Every piece of data is stored / represented in the above mentioned combination, which would be referred to as neural signature for a given data set. Likewise, any unit from the metric or the numbering system can be employed to code the information or the data. With change in the coding system used, the number and combination of the unique digital neurons created would also vary substantially.
The information is stored in the form of the neural signature. The information or the data is a letter, a word or even sentence. However the information coded is not limited to the alphabetical alignment, it can even be an image which is HEX coded to give a neural signature. When the same information or the data is perceived at the next instant, the neural signature pertaining to that information is triggered thus working like the memory and the mind of the human brain.

Object of the present invention
The present invention objects to provide the artificial neural network that mimics the biological neural network of the brain.
Another object of the present invention is to provide the digital neurons that are capable of creating the memory patterns using any unit of metric system for coding such as HEX coding, octal system, binary system, cryptographic hash.
Yet another object of the present invention is to provide the system for storing any type of information in the form of neural signature.

Detailed description of the Invention
The present invention pertains to the digital neurons such that these neurons create a memory patterns for any type of the information or the data using any unit of metric system for coding such as Hex code system, binary system, octal system, cryptographic hash and like
Considering the HEX coding system wherein the Hex code forms the neural signatures which are then stored as the memory like the brain. HEX value is an8 digit code (4 hex code till 255) starting from 00000000 to ffffffff. Each hex code is (with decimal value from 0 to 255) represented by first two digits such that first two digits are is one neuron. So the example mentioned above, we could have 4,228,250,625 unique neurons (more than 4 billion neurons) and their combinations. Thus any information perceived can be HEX coded without any chance of repetition of the code. The HEX coding can change to any other unit and total digit code can change from 8 to 16 or even more than 16 and less than 8. This creates a scenario as in actual brain where data is stored. Every piece of data is stored / represented in the above mentioned combination, which would be referred to as neural signature for a given data set. Data storage is very efficient and a lot of data can be associated with each other using neural signatures. These neural signature can be saved digitally in a file format with other associated neural signature data in it or in a directory format with other associated neural signature data in a file format representing memory. The information is not just alphabetical information like letter, word or a sentence, but it can be an image or even an incident or the communication as well.
With Natural Language Processing (NLP) the above method can be used for storing any language data, which gets converted into neural signatures which can have its own properties. The same methodology can be used as a storage engine for any database requirement where the stored reference also needs to be processed. There can be combinations of neural signatures possible for various memories. E.g. ‘You are a Good Boy’ where You = 00000005, are = 00000006, a = 00000007, good = 00000008 and boy = 000000009. Here the sentence in memory would be 00000005000000060000000700000008000000009 (You are a Good Boy) and then the same can be further optionally mapped, if required to different unique neural signature like 00000010. So when ‘You are a good boy’ is in input, neural memory of 00000005000000060000000700000008000000009 is activated and inturn 00000010 is activated. This allows us to create a brain with dataset, as being stored in the brain. These dataset are referred to as memories and each unique neural signature can contain large combination of memories. Such memories can perform action like taller, which in turn performs a function with two or many arguments within. Actions like walk can be activated by the neural signature and the action programmed in the action of the activated neural signature can be performed. This way of storing and programming the memory with actions/logic would revolutionize the future of AI. A main advantage of such a learning process would be enabling the AI to learn by observing and not by continuous training. The AI will make mistakes as human, but will keep learning (unsupervised) with the resources in his circle of trust (emotion). The above architecture helps us to create an actual working mind. There would be different algorithms for conditional programming of each neural signature based on the action it has to perform like thinking, dreaming, running, talking or walking. The neural signature will be able to store patterns in memory like human faces, which would be activated on viewing the same human face again, provided camera and other hardware technology is incorporated with AI
Example 1
Suppose the data or the information is the image of a girl described as XYZ is a beautiful girl. Now the system of the present invention forms one neural signature for the name of girl which is XYZ, now this neural signature will have two associated neural signature assigned to 'beautiful' and the 'image' of the girl provided which can be referred to as child neural signature. This together constitutes one file of a memory. When the name is perceived at the next instant, the neural signature for the 'beautiful' and the 'image' are simultaneous triggered firing a complete memory. These neural signature can be saved digitally in a file format with other associated neural signature data in it or in a directory format with other associated neural signature data in a file format representing memory.
,CLAIMS:We claim:
1. An artificial neural network comprising the digital neurons to create memory patterns for any type of information or data using the metric system for coding such as HEX coding, octal system, binary system, cryptographic hash.
2. An artificial neural network as claimed in claim 1 wherein, the system for storing any type of information in the form of neural signature.
3. An artificial neural network as claimed in claim 1 wherein, in the HEX coding system, the Hex code forms the neural signatures which are then stored as the memory like the brain.
4. An artificial neural network as claimed in claim 1 wherein, any information perceived can be HEX coded without any chance of repetition of the code.
5. An artificial neural network as claimed in claim 1 wherein, the HEX coding can change to any other unit and total digit code can change from 8 to 16 or even more than 16 and less than 8, which creates a scenario as in actual brain where data is stored.
6. An artificial neural network as claimed in claim 5 wherein, every piece of data which is stored or represented using HEX code is referred to as neural signature for a given data set.
7. An artificial neural network as claimed in claim 1 wherein, the neural signature can be saved digitally in a file format with other associated neural signature data in it or in a directory format with other associated neural signature data in a file format representing memory.
8. An artificial neural network as claimed in claim 1 wherein, the neural signature will be able to store patterns in memory like human faces, which would be activated on viewing the same human face again, provided camera and other hardware technology is incorporated with AI.
9. An artificial neural network as claimed in claim 1 wherein, the information stored in the neural signature can be in alphabetical information like letter, word or a sentence or an image or even an incident or the communication.
10. An artificial neural network as claimed in claim 1 wherein, the Natural Language Processing (NLP) can be used along with the HEX method for storing data in any language which gets converted into neural signatures having own properties.
11. An artificial neural network as claimed in claim 1 wherein, there method allows to create a brain with dataset, where the dataset are referred to as memories and each unique neural signature can contain large combination of memories.
12. An artificial neural network as claimed in claim 1 wherein, the system contains different algorithms for conditional programming of each neural signature based on the action which has to be performed, for example: thinking, dreaming, running, talking or walking.