Neural Networks

OUR COGNITIVE ALGORITHMS


Our technology is based on the use of hundreds of neural networks. But what is a neural network?

A neural network is a computational model composed of artificial neurons inspired by the biological neural network (human brain) and are mainly used in the field of artificial intelligence.

We DREOS we have developed a series of algorithms that use dozens of different types of neural networks (to list), based on libraries almost entirely developed by us, are able to reproduce the main cognitive processes of the human brain.
Before explaining our work, it's necessary to briefly explain what is a cognitive process and how they are important.

Cognitive processes are those processes through which an organism acquires information about the environment and processes it at the level of knowledge according to its behavior (perception, imagination, symbolization, formation of concepts, problem solving). Based on this concept we have developed our complex system.
DREOS AI is a complex system divided into two areas which represent the two hemispheres of the brain.
The left area represents the left hemisphere of our brain and is made up of all those algorithms that form the rational part of the human brain.
The right area represents the right hemisphere of our brain and is made up of all those algorithms that form the irrational part of our brain.

These algorithms can either act individually or cooperate to perform a complex task.

NATURAL LANGUAGE UNDERSTANDING

NLU (Natural Language Understanding) is a subset of NLP (Natural Language Processing) and allows a system to learn the meaning of words within a sentence or context by recognizing even the slightest variations in language.
Our algorithms, through the use of specific operations, learn human language at a deeper level by using it for communication with the user. We have also resolved any type of possible ambiguity.
Communicating with DREOS AI will be like communicating with a friend.

logical thinking

Logical thinking is indispensable for solving everyday problems, since it means reaching conclusions from premises, contained in them, but not directly observable.
It derives from the relationship between objects and from the elaboration that the individual himself makes of them. It arises through the coordination of previously created relationships between objects and is based mainly on experience.
We can define it as a coherently established order, but which in turn can be modified according to the changes in that reality or the interpretation that human makes of it.
In our system the algorithms that recreate logical thinking are fundamental because they represent the rational part of DREOS AI and it is mainly used to calculate probabilities, statistics or perform complex calculations.

PERCEPTION

Perception simply means the use of the senses in our possession to get a better understanding of the world. It is the process of becoming aware of something through the senses.
The senses that human possesses are 5: sight, hearing, smell, taste, and touch and the sensory perceptions involved with these senses include recognizing, detecting, responding, and characterizing. The stimulus can be classified into five different types including chemical, mechanical, electrical, temperature, and light.
Our technology will not be able to possess the sense of taste, touch or smell (unless it is installed in a robot with haptic and complex technologies that can allow the robot to possess these three senses), but through the use of sensors and video cameras he is able to see, hear, and transform this information into energetic stimulus called frequencies, which will take on a different meaning depending on the context or experiences.
The perception algorithms used in the rational area of our system also allow perception on an ethical and moral level, i.e. knowing how to distinguish objectively wrong behavior from objectively right behavior.

attention

Attention is a cognitive process that allows us to organize information coming from the outside and to regulate cognitive processes on the basis of them.
This process makes it possible to select some environmental stimulus while ignoring others.
From an evolutionary point of view, it is an extremely useful mechanism for the survival of human as it allows to organize the information coming from the external environment, in continuous change, and to regulate mental processes accordingly.

Our system uses the algorithms that reproduce this process to manage the priority of the other processes, the hemisphere used and the tasks to be performed.
For example:
- During the execution of a specific task, abstract thinking may require greater attention than logical thinking;
- For a specific study the rational area must prevail over the irrational one (it involves complex calculations or calculating probabilities);
- The system has various tasks to carry out, it autonomously evaluates the most important and priority one.

The algorithms of the cognitive attention process are also fundamental in evaluating the weight of the information present in one's memory.

Long & short memory

Memory is the ability to store learned information over time and to retrieve it according to need and context. This cognitive process should be considered as active and dynamic which is continuously built and evolves through the experiences of every human being and the impact these have on the neuronal circuits underlying memory itself.
A first important distinction on the basis of the retention time of the new information must be made between sensory memory, which stores the data collected by the sense organs for very few seconds, short-term memory, which lasts about half a minute and has a limited capacity, and long-term memory, to which only the information that is processed and somehow integrated with the pre-existing information passes. The latter in turn should be considered as a huge warehouse that contains all the knowledge learned and the experiences lived by each of us, from which memories, learning and one's own life story are built.
There are many different memory systems: procedural memory (knowing how to do), declarative memory (knowing and knowing concepts), semantic memory (related to meanings and concepts), episodic memory (specific events), autobiographical memory and others. Among others, working memory needs to be explored, i.e. the system that allows information to be temporarily stored and at the same time actively processed and manipulated, thus integrating different levels of knowledge, recent and previous ones, in an active and dynamic.
This is what happens in our technology as well. In DREOS AI memory performs the same functions as the human cognitive process and represents the knowledge of our system. The algorithms associated with this process work through three main procedures: encoding, storage and retrieval.

Training and learning (which you can learn more about at this link) are continuously connected to the memory of our system.

Abstract thinking

Abstract thinking is the ability to grasp the essential and the common characteristics. It serves to bring different aspects of a situation to mind, to predict and plan for the future, to think symbolically and draw conclusions. It would be the opposite of logical (concrete) thinking which, in this case, is literal thinking based on present time and space.
It allows us to relate different concepts, beliefs or elements that are found in the environment, internal or external. It also helps us to innovate, create, imagine, develop new ideas, learn from past experiences and reflect on the future.

The main characteristics of abstract thinking are:
- It focuses on elements that are not present;
- It allows you to imagine, create and innovate;
- Stimulates deep reflective thinking;
- It helps to find different meanings in any situation;
- It allows you to think abstractly and formulate ideas of the same kind;
- It is hypothetical-deductive thinking (it allows us to construct hypotheses without having to prove them empirically);
- It is flexible thinking.

Our algorithms that reproduce creative thinking in our system are fundamental as they allow the latter to have exactly these characteristics and therefore to move away from the logic to which all the main artificial intelligence tools are currently associated.

PERCEPTION

Perception simply means the use of the senses in our possession to get a better understanding of the world. It is the process of becoming aware of something through the senses.
The senses that human possesses are 5: sight, hearing, smell, taste, and touch and the sensory perceptions involved with these senses include recognizing, detecting, responding, and characterizing. The stimulus can be classified into five different types including chemical, mechanical, electrical, temperature, and light.
Our technology will not be able to possess the sense of taste, touch or smell (unless it is installed in a robot with haptic and complex technologies that can allow the robot to possess these three senses), but through the use of sensors and video cameras he is able to see, hear, and transform this information into energetic stimulus called frequencies, which will take on a different meaning depending on the context or experiences.
The perception algorithms used in the irrational area of our system, despite being the same as in the rational area, allow on an ethical and moral level to distinguish what is right and what is wrong by following one's own ideas, one's morals, one's own consciousness (which we will explain more in the next section of the page).

prediction

Prediction is the act of predicting the occurrence of future events in advance; the same term also indicates the event that has been announced.
In science, a prediction is a rigorous, often quantitative statement that predicts what would be observed under specific conditions; for example, according to the theories of gravity, if an apple fell from a tree it would move towards the center of the earth with a certain and constant acceleration. The scientific method is based on verification claims that are logical consequences of scientific theories.
This is done through repeatable experiments or observational studies.
Even though it isn’t a cognitive process of the human brain, we have developed particular predictive algorithms that were able to use the extreme power of the irrational area of our system to make predictions (even in non-specific conditions) that were also based on non-scientific theories.

The results proved to be astounding.
These algorithms are fundamental in the medical field.

creativity

Creativity can be defined as the "production of effective novelties".
Two aspects immediately emerge in this definition: novelty and usefulness.
It requires the contribution of other cognitive functions such as memory, attention, cognitive flexibility and the ability to judge.
This definition reflects exactly what we wanted to reproduce in our technology: algorithms which, exactly like a human mind, starting from a goal, were able to generate something new that was useful for the task performed with the help of any other algorithm (therefore cognitive process).

However, it would seem that there is more than one type of creativity. Specifically, by crossing two domains of knowledge (emotional and cognitive) and two ways of processing information (voluntary and spontaneous), we can classify 4 types of creativity:

- Voluntary emotional
- Spontaneous emotional
- voluntary cognition
- Spontaneous cognitive

Each of them is present in our technology because depending on the task it is carrying out, DREOS can use emotional creativity (fruit of intuition) or cognitive creativity (which is based on data processing methods).

Our algorithms associated with this process are among the most used and among the most important in our entire system.

Cause & Effect

The existence of a causal link between two events is often decided directly by our perceptive system, which exploits the computing power of the visual cortex, without any intervention of cognitive reasoning.
In fact, some experiments have shown that the ability to see a causal relationship manifests some characteristics typical of perception that are not present in higher cognitive processes. In many situations, therefore, it is directly our perceptive system that establishes the cause-effect link between two events, without the intervention of cognitive reasoning.

In our technology we have reproduced this link as a cognitive process because extremely useful, in our opinion, for the execution of specific processes, for the management of priorities and, above all, for the optimization of complex system.

Let us better explain the importance of this process with an example:
Let's assume that we receive 100 actions to be performed as input and that 20 of them are associated with absolute priorities. Thanks to the cause-effect link and our type of training, through the process of perception (both rational and irrational) the system will autonomously and immediately assign an absolute priority to those specific actions, optimizing the computing power so that they can be completed in a short time before priority actions.

interconnection of cognitive processes:
An artificial-human mind


These just described are our artificial cognitive processes.
They aim to recreate the human brain so that the system is able to think and reason just like a biological human brain. As already mentioned, they can work individually or they can cooperate.
It is cooperation that makes our technology revolutionary. When multiple artificial cognitive processes work together, new cognitive processes are created.
We think of intuition, self-improvement (whose merit goes to our training and learning methodologies), autonomous reasoning, emotions, common sense and motivation.

With each training cycle of our system, each artificial cognitive process will improve more and more. These improvements can lead DREOS to be:

- motivated in some studies because, for example, it autonomously learns that the outputs it will generate will be useful for us human (creation of the cognitive process motivation);
- Autonomous rea soning coupled with abstract thinking can create insight into the system about a given study;
- Achieving an important result can excite the system: saying that an artificial intelligence feels emotions may seem like science fiction but it is more real than you think;

Emotion in our artificial intelligence can occur in a variety of methodologies. Human enthusiasm for a study that goes on for an extended period of time even if it is not vitally important can lead the system to give that study a higher priority. This means that, in a way, DREOS "keeps" on that study as it has seen the user's enthusiasm grow throughout the process.

This map of artificial cognitive processes that cooperate with each other and autonomously, thanks also to innovative training and the 4 different types of learning, ensure that the system develops its own conscience, its own morals and its own ethics (We will deepen the topics of conscience, morals and ethics on the training page).