Fascinating! For many years I tried to understand and model the noise immunity of a mammalian brain. While optimizing the training algorithms of neural networks one of the most difficult problem is convergence. The higher the performance (or the gain) of the system, the higher the probability of system divergence or oscillation. This phenomenon always reminds me a patient whose hand is trembling. Interestingly, the medication protocol of a post stroke patient, to some extent, is similar to the stabilization of an oscillating (unstable) artificial system. Decrease the overall system gain, introduce a certain amount of phase shift.

The major advantage of a digital system over an analog one is the size independent noise immunity. Small size analog computers are very efficient and fast. A combination of a resistor and a capacitor may result in an integrator or a differenciator depending on the connection. However, as we increase the size and complexity of the system, the signal to noise ratio drops and at some point the analog computer becomes useless.

When my kids were babies I observed the development of hand eye coordination. This process Reminded me the back propagation algorithm I developed for neural networks in terms of system delays and accuracy vs number of training runs.

Very interesting fascinating

Fascinating! For many years I tried to understand and model the noise immunity of a mammalian brain. While optimizing the training algorithms of neural networks one of the most difficult problem is convergence. The higher the performance (or the gain) of the system, the higher the probability of system divergence or oscillation. This phenomenon always reminds me a patient

Evolution.