EvolutionaryDNN

Plan

1. Implement CNN in TensorFlow

• Understand of CNN (Mathematical details)

2. Identify adjustable parameters

• Feature maps, weighting

3. Pick / Survey 1 to 2 interesting Problem

• Accuracy around 80%

4. Implement evolution of parameters

• Differential evolution

5. (Optional) Implement evolution of strategies

• Speed up / predict the performance of DNN without evaluation

6. Evaluate the final DNN and short report