Bayesian inference for Scala
What
- Build a probabilistic simulation in idiomatic, immutable Scala
- Match the simulation with real-world observations
- Rainier will infer the parameters for your simulation that could lead to that data
- Make predictions from those parameters to optimize your decision-making
How
- Fixed-structure, continuous-parameter models like Stan or PyMC3
- Custom TensorFlow-like computation graph with auto-diff
- Gradient-based inference with Hamiltonian Monte Carlo
- Dynamic compilation to low-level bytecode for speed
- Pure JVM operation for ease of production deployment