While playing with evolutionary algorithms, I had the startling realization that all genetic mutations are bad. It’s common knowledge that biology abhors genetic mutation, and I assumed that was only because mutations cause cancer. But my computer programs are immune to cancer, and they also abhor mutations. This is counterintuitive, given that evolution requires mutations to procede.
For proof of the fact that mutations are bad, consider that evolution is an optimization algorithm, and after it reaches a local optimum further mutations will be strictly detrimental. The concept of evolutionary pressure is the ability of an evolutionary algorithm to remove deleterious mutations from a population. If mutations accumulate faster than they can be removed then the population will suffer a genetic collapse. This is a common failure mode of evolutionary algorithms.
The ideal evolutionary algorithm would have at most one mutation in each individual, and each of their lives would be an experiment to evaluate that single mutation. And then through many generations of chromosomal crossover the best mutations would combine into a single genome.
You might be rediscovering Fisher’s geometric model. A refinement to your current model you could consider is that close to, but not exactly at, the local optima, sufficiently small mutations have a 50% chance of being beneficial.
While playing with evolutionary algorithms, I had the startling realization that all genetic mutations are bad. It’s common knowledge that biology abhors genetic mutation, and I assumed that was only because mutations cause cancer. But my computer programs are immune to cancer, and they also abhor mutations. This is counterintuitive, given that evolution requires mutations to procede.
For proof of the fact that mutations are bad, consider that evolution is an optimization algorithm, and after it reaches a local optimum further mutations will be strictly detrimental. The concept of evolutionary pressure is the ability of an evolutionary algorithm to remove deleterious mutations from a population. If mutations accumulate faster than they can be removed then the population will suffer a genetic collapse. This is a common failure mode of evolutionary algorithms.
The ideal evolutionary algorithm would have at most one mutation in each individual, and each of their lives would be an experiment to evaluate that single mutation. And then through many generations of chromosomal crossover the best mutations would combine into a single genome.
You might be rediscovering Fisher’s geometric model. A refinement to your current model you could consider is that close to, but not exactly at, the local optima, sufficiently small mutations have a 50% chance of being beneficial.
Thank you