At the moment, the proof-of-concept project has the following features:
- 100% Neurons
- Firing correctly
- Weights interpreted correctly
- 100% Neural Networks
- Firing across a network
- Sane results
- 99% Genetic Algorithms
- Breeding Neural_Networks using Bitstrings of weight arrays
- Breeding Neural_Networks using vector<float> weight arrays
- Both crossover and mutation working perfectly (check for floats)
- All that's left is to tweak and check
- 99% Agents
- Wrapping Neural_Networks perfectly
- All that's left is to tweak and check
- 99% World
- Easily creates arbitrary numbers of agents
- Can run simple simulations of network firing and Agent breeding
- All that's left is to tweak and check