Structure

• data-adaptation - this folder contains script for data preprocessing and columns filtering
• py - all the sources of python code including notebooks
  • model
    • autoencoder - source code of LSTM autoencoder
    • stan - all the notebooks of experiments with Stan
  • sim - not relevant
• robot-data
  • new_data - data received from Ronens student
    • normal - normal data of 3 different simulations
    • test - data contains anomaly for different simulations
• src - old Julia code, not relevant

Installations

For running notebooks need to install next packages:

• pystan
• pandas
• scipy
• heapq
• tqdm
• sklearn
• pickle
• keras
• tensorflow

Running

All the experiments with the data are in py/model/stan directory

• 2D-GMM_pick_miss_cup - Multidimentional GMM with data of picking cup (Autoencoder and GMM)
• ProbabilisticModelExperiments - Experiments with old data, not relevant
• ProbabilisticModelExperiments_build - Experiments with new data of moving robot (PCA and GMM)
• ProbabilisticModelExperiments_cans - Experiments with new data (PCA and GMM)
• ProbabilisticModelExperiments_corr - Experiments with new data (PCA and GMM)
• ProbabilisticModelExperiments_pick_miss_cup - Experiments with new data (PCA and GMM)
• StanAutoencoder_pick_miss_cup - GMM with data of picking cup experiments (Autoencoder and GMM)