NeuralEngine
A Game Engine with embeded Machine Learning algorithms based on Gaussian Processes.

This project combines 3D real time engine with machine learning algorithms based on Gaussian Processes sparse approximations techniques with optimization via Expectation Propagation (EP). This probabilictic framework enables deep neural architectures in continous function domain. With this approach no data for prediction, like for other non-parametric models, is needed anymore. This enables batch learning and increases scaleability. The probabilistic framework is implemented for optimization on graphics card and allows generation of character animation, interaction with users or probabilistic mesh deformation. The project is developed by XXX member of XXX.

Installation

  1. Install Visual Studio 2019 with C++ builders.
  2. Run 'RunMeFirst' command to install all prequisits.
  3. Open build/NeuralEngine.sln file.
  4. Enjoy!

Advise for building own projects

The example projects depending on a bunch of libraries, where all needed *.dlls are copied to the execution folder during build. The easiest way to create an own project:

  1. Copy an existing example project from samples\MachineLearning.
  2. Paste the copy into projects\MachineLearning.
  3. Rename the project folder, e.g. "MyAwsomeGPProject".
  4. Go inside your copied folder and rename the main-cpp, .vcxproj, .vcxproj.user to "MyAwsomeGPProject".*.
  5. Inside Visual Studios Solution Explorer right click on Projects/MachineLearning folder and Add->Existing Project to import.
  6. Rename the project inside Solution Explorer again.
  7. Use the projects main-cpp to start your project.