using gradient free oprimization algorithms in machine learning
- Genetic algorithms are stochastic search algorithms
- inspired by the process of natural selection
1, two forward propagations one for the parent and the other for the child
2, the child weights are mutated by a random value
3, compute the loss function of the child and the parent
4, if the child loss is less than the parent it will update the weights
- Repete this until the model converges
