What’s Really Going On in Machine Learning? Some Minimal Models
A fascinating research report from a fascinating researcher. Wolfram studies cellular automata, which are very small and simple models of abstract computation in which each “cell” adjusts its state according to the state of some number of neighbors. These automata are even simpler than neural networks, but they, too, can learn. It appears that machine learning, in this case at least and perhaps in all cases, including animals and people, may be a matter of sampling lots of abstract computations and refining those that seem to have some slight utility for addressing the current situation.