A guide for what switching is and how it’s used
Switching is simply a switch in behaviour of something. Let that be a light switch. Or let that be a switch in behaviour of a gas , or change in DNA. The main idea of switching comes from Epigenetics which from the Google definition is as follows: “Epigenetics is the study of how your behaviors and environment can cause changes that affect the way your genes work. Unlike genetic changes, epigenetic changes are reversible and do not change your DNA sequence, but they can change how your body reads a DNA sequence.”
Now imagine, having a similar epigenetic behavior but in simulated beings or algorithms. This is where we get our idea of Switching from. Since like epigenetics our Reversal’s are NOT mutation operators, BUT rather they cause a change in the environment! How? By forcing the algorithm to explore a different search space altogether (a change in environment). As a result of such changes, after coming back to the original environment, it might explore a different route and not get stuck in local minima (having escaped it by backtracking).
Another way to think of it is in the terms of Backtracking. Our algorithm basically backtracks up the loss landscape to find a better route, and then comes down the loss landscape.
