I was curios if anyone would be interested in discussing their thoughts on how to best design your features for Gen. Optm. I have had a few different builds and thoughts now and have some questions.
Would a genetic Optimization be best fit for parameters that have less "step counts" or options per parameter.... for example 4 or 5 options per parameter vs 100 or 1000 options? The optimization isnt using gradient descent is it? It just looking for what inputs lead to higher fit values?
all equal which of these parameter options would you choose for a genetic optimization?
all of these have around 10,000,000 options
10 parameters with 5 Step couts or option,
7 parameters with 10 Steps,
6 parameters with 15 Steps, *Edit
3 with 200 *Edit
Hopefully that gets the point across. Just curious everyone's thoughts.
feature design for Genetic Optimization
Questions about MultiCharts and user contributed studies.
3 posts • Page 1 of 1