(11-08-2024, 04:45 AM)z-s Wrote: Yes as first step and may could recognise pattern and images after it
(translated by Google translator)
The information I have says that you have two ways to achieve your goal.
The first way is the classical one.
First you have to choose a classical neural network structure that can do the job you need.
Then you will need to prepare data to train this neural network.
Then the neural network must learn from this data.
If the neural network cannot learn, then you will be forced to change the structure of the neural network and try to train it.
If you are not a specialist in classical neural networks, then it will be difficult for you to do this kind of work correctly.
It is for this case that I created for myself SB-neurons, which I can consciously collect into a SB-neural network.
In this case, the structure of this SB neural network will be formed automatically, in accordance with the task that each SB neuron will perform.
Creating an SB neural network requires less work than preparing data for training a classical neural network and training it.
But, to create a SB neural network, the developer must clearly understand what work each SB neuron should perform.