More than 10 years ago I got interested in the idea of Artificial Neural Networks.
Initially it really expressed me as it looked like a possibility of modeling animals'/humans' brains!
After some time I got a feeling that:
- I had not enough math/logic/English(my native language is Russian) skills to read and understand most of the literature on the topic.
- I had no super skills in programming to express/implement any ideas without overcomplicating it and spending too much time to just test an idea.
- I had no feeling that the information and ideas that I had found gave me a feeling that I could model a brain. It looked like all Artificial Neural Networks were about modeling some very limited subtask of a part of a brain. Some of them were really just using the same concepts for a completely different tasks that are useful for human but do not have anything common with what happens in brains.
Almost at the same time I got interested in a different area: genetic algorithms. Again I was not interested in application of it for real-world problems.
What I wanted was to use it to do evoltion of something living in my computer :)
At this moment I came to an idea of applying Genetic Algorithms to Neural Networks - to build the structure of it and find the weights using evolution.