세부사항
설명
리뷰
질문
NeuroEvolve is a plugin that allows the user to create a custom fully connected neural network and train it using genetic algorithm. User only needs to know blueprints and can easily implement the functionality of the plugin on already existent blueprint actors. Every generation can be saved and used as starting point later. When finished with training user can take the best performing weights of the neural network and use them for in-game agents.
Video examples (not a tutorial): https://youtu.be/dW4kz5IWa0Q https://youtu.be/MPIb6OsUZzQ
기술적 세부사항
Features:
- Ability to create sequential neural networks and train them using genetic algorithm
- Best performing netoworks can be used in packaged games to add logic to in-game agents
Code Modules:
- Name: NeuroEvolve; Type: Runtime; LoadingPhase: Default
Number of Blueprints: 9
Number of C++ Classes: 9
Network Replicated: No
Supported Development Platforms: Windows
Supported Target Build Platforms: Windows
Documentation: Link included with the plugin
Example Project: https://drive.google.com/file/d/1TbsHhOaED-Fu0gSL5dq4xIZ1BB5uBE4f/view?usp=sharing
Important/Additional Notes:
관련 콘텐츠
