PROGRAMMING IA

Pyton /Linux /OPENcv /TensorFLOW /Caffe /ImageNET  /DeepNeuralNetwork

It is at the Software level that the idea of ​​Artificial Intelligence is born through deep neural networks and a way of programming through entry and re-entry of data, comparisons, recomparations and finally a response that will help us make decisions. The input data from images, sounds or other sensors in the environment are stored in nodes that we will call neurons. All this information stored in the neurons will interact according to how the algorithm is designed allowing an output or response that could be used to perform an action such as converse, move, act, turn on, turn off, interact with human beings or other robots with the same capacity. In the beginning our Droid will contain a set of information of images, sounds or other data coming from global databases like IMAGENET. Once the software is started in the Droid, it will begin to receive inputs from its own sensors, such as images from the cameras or measurement sensors that it has incorporated, thus giving way to the beginning of its learning while it is alive or on. The more information you enter into your system the richer it will be in the quality of your answers and therefore in the quality of your decisions. Equal droids located in different geographical locations will improve their capacity little by little in their respective environments making them then more capable in their own environment where they were turned on Learning algorithms put in place to look through cameras and listen through microphones They have already far exceeded man's ability to recognize the environment that surrounds him. This is just beginning, thousands of AI software are running right now and we are beginning to see interesting results in all areas. Softwares that paint, compose music, recognize faces, communicate with each other. The greatest art will be developed by programmers with sufficient sensitivity to configure the neural network as efficiently as possible in terms of prediction of the medium. Analogo to the Human Brain once learned, the channels are established for each type of response. A self-contained car will then require you to drive it manually in a certain environment before letting it drive alone. It can be complex to enter each of these disciplines and in fact it is, but the results are promising

To start programming, first of all we need a computer that has enough processing power to run a lot of images or results of sensors in real time in order to begin to dismember and analyze this information in different ways depending on how is the analysis algorithm devised. We will also require an open operating system such as Linux that allows us to communicate the processor with the different sensors, such as cameras or other types. Then we will get a graphical processing computer of the GPU type that has control inputs and outputs to connect the sensors. This computer will work as a companion to the main computer, be it a Drone, an Automobile or a PC. We will install the operating system in this case in a NVIDIA Texon TX2 brand integration computer that will allow us to load in the algorithm a visual recognition of characters such as numbers, or parts of a person's face, or retina, or we can also load an algorithm that identifies the transit lines to implement a self-contained car. The good thing is that we do not have to start creating the algorithms from scratch because they are already created and there are thousands available to start training. Also come pre-trained then just run these work, so the job will be to refine the way they deliver the results.