Skip to content

Latest commit

 

History

History
33 lines (28 loc) · 1.39 KB

SCENARIOS.md

File metadata and controls

33 lines (28 loc) · 1.39 KB
Raw

Scenario Descriptions

Scenario 1

Files used:

  • s1_simulate.py
  • s1_config.py

This scenario is used for a simple simulation run to test the various settings. One can try out different types of generated data (DPM, MFM, GM) as input to the algorithm and also try different initialization (see config file for the various settings and their description). The respective config file (s1_config.py) stores all parameters that can be changed and also serves as a documentation of all the settings that can be made.

Scenario 2

Files used:

  • s2_simulate.py
  • s2_config.py

This scenario is used to simulate a simple object tracking scenario. The data is generated by a DPM with parameters given by the config (s2_config.py), where also noise is included in the process. Out of the noisy measurements $y=x+y$ the true object positions $x$ are tried to be recovered with an MMSE estimator where a clustering for y is done beforehand with VI. The simulation is performed for multiple times for each object count N and the average MMSE result including confidence interval is used to compare to sampling techniques.

Scenario 3

Files used:

  • s3_simulate.py
  • s3_simulate_Nvec.py
  • s3_config.py Clustering scenario where various metrics such as the accuracy score and OSPA are computed at the end of the simulation run. We use this scenario to compare results of variational inference using a MFM model vs. using a DPM model.