In Chapter 4, Deep Learning for IoT, we learned about different DL models, such as MLP, CNN, RNN, and so on. Now, we will see how we can use genetic algorithms with these DL models. Genetic algorithms can be used to find the optimized weights and biases, and people have tried them. But the most common use of genetic algorithms in DL models has been to find optimum hyperparameters.
Here, we use genetic algorithms to find the optimum CNN architecture. The solution here is based on the paper Genetic CNN by Lingxi Xie and Alan Yuille (https://arxiv.org/abs/1703.01513). The first step will be finding the right representation of the problem. The authors presented a binary string representation for the network ...
