From a discussion I dad with "the guy from McGill", about neuronal networks as a support for my theory.
Sensors are invading the world. Deep learning has became popular as neuronal networks have met big data. When that data is about the physical world we get to epistemology (we'll transcend engineering). In fact, there is a strong relation between classical science and engineering. Engineering is seen as the application of science, but science and engineering are both plagued by anthropocentrism and the desire to get closer to god. Science is about learning how stuff works, believing that we can get to know how stuff ''really works''. Engineering is about making stuff work for us to solve a problem or satisfy a need, assuming that we do know how stuff ''really works''.
It is interesting that now we see a shift, a departure from trying to understand. There is an acceptance that we don't really need to understand, which is actually fundamental.
We can't reverse evolution, we can't understand who things really work, how things became the way they are.
See the video: Deep Learning: Intelligence from Big Data: http://youtu.be/F3v0nTTs7O0
Also see Steven's intervention during RIW gathering - language is action, see also Searl. Cognitive processes are all related to action, also related to pragmatism. Knowledge is about doing something, patterns of knowledge emerge in action.
ToDo: See Geoaf Hinten, prof in Toronto, and Yann LeCunn in Montreal, in deep learning, contact him with my theory in epistemology.
See also about hidden variables in neuronal networks. There are inputs, outputs, and hidden variables.
In the video, Steve said that we don't really understand what goes in neuronal networks. That is probably because we don't know how to think about it. Perhaps my theory can help. People probably still approach it in the classical way. There needs to be a new underlying epistemological theory missing there, see if I can provide the solution... For that I need to understand how people think and talk about machine learning, deep learning, and see how that compares with the old epistemology, detect tendencies to brake away.
It also seems that there is human bios in neuronal networks, since the punishment and the reward is a human input. The goal of the neuronal network seems to be training of the network to recognize a pattern, and after to use that configured network to recognize the same pattern elsewhere. But in reality, biological systems get rewarded and punished by nature, catch pray, eat, die, etc. So the structure of the neuronal network emerges in relation with nature, and there is probably something in there with memetics, because these memes that ''need'' to reproduce will carry further the good pattern and the bad ones will just disappear, but why these things that ''want''to replicate. So look into memetics as well, the might be something there.
One lead for further development is the free energy principle.