DeepNotebooks is an automated statistical analysis tool build on top of SPNs. They are currently being developed by Claas Völcker at the ML group at TU Darmstadt under the supervision of Alejandro Molina and Kristian Kersting.
DeepNotebooks are a tool to automatically analyze your datasets using powerful graphical models for inference. They are build to provide fast access to your data and summarize relevant statistical information directly without fine tuning. Due to their structure, they also expose the underlying models for a deeper analysis.
SPFlow is the underling inference library powering the DeepNotebooks. It is developed at the ML group in Darmstadt.
GitHub: https://github.com/alejandromolinaml/SPFlow
Website: https://ml-research.github.io/
- Völcker, Claas "Interactive data analysis using Sum-Product Networks", bachelor's thesis at TU Darmstadt, 2018, available here: https://userdata.d120.de/cvoelcker/thesis.pdf
- Claas Völcker, Alejandro Molina, Johannes Neumann, Dirk Westermann and Kristian Kersting. "DeepNotebooks: Deep Probabilistic Models Construct Python Notebooks for Reporting Datasets" In "Automating Data Science 2019"
If you find DeepNotebooks useful, please cite us in your work:
@incollection{
voelker2019ads,
year = { 2019 },
crossref = { https://github.com/cvoelcker/DeepNotebooks },
title = { DeepNotebooks: Deep Probabilistic Models Construct Python Notebooks for Reporting Datasets },
pages = { },
booktitle = { Working Notes of the ECML PKDD 2019 Workshop on Automating Data Science (ADS) },
author = { Claas Voelcker and Alejandro Molina and Johannes Neumann and Dirk Westermann and and Kristian Kersting } }