# Problem Would be nice to be able to actually search through possible *autoencoder architectures* without knowing specifics of what it should or should not be. # Solution An *algorithm* that can essentially *"build"* and *"evaluate"* different architectures (e.g. NEAT, WANN). ## References + https://github.com/iamgroot42/NEAT-Keras + https://github.com/CodeReclaimers/neat-python + https://weightagnostic.github.io/ + https://arxiv.org/abs/1906.04358