We describe the design of Omnigram Explorer (OMG), an open-source tool for the interactive exploration of relationships between variables in Bayesian networks (and other complex systems). OMG is designed to help researchers gain a holistic, qualitative understanding of the relationships between variables, specifically providing interactive, visual support for observational sensitivity analysis. OMG is especially useful for high-lighting dependencies between variables and small groups of variables. It's designed for exploratory analysis of BNs (and other models) and for communicating the salient features of models to non-specialists. Kevin Korb - http://www.csse.monash.edu.au/~korb Tim Taylor - http://www.tim-taylor.com/ Seventh Annual Conference of the Australasian Bayesian Network Modelling Society conference (ABNMS2015) - Monash Caulfield, Melbourne, Australia 2015: http://abnms.org/conferences/abnms2015/

1. 1
Omnigram Explorer:
A New Interactive Tool for Exploring Bayesian Networks
Kevin B Korb1 & Tim Taylor2
1Faculty of Information Technology, Monash University
2University of London International Programmes
www.tim-taylor.com/omnigram
Australasian Bayesian Network Modelling Society 2015

2. Omnigram Explorer
1. Motivation
2. Omnigram Explorer Tools
a. Concepts
b. Single Node Brushing
c. Multi Node Brushing
d. Linked Brushing
e. Flash Mode
3. Exploring Dependencies
a. Viewing D-Separation
b. Explaining Away
c. The Best Explanation
4. Conclusion

3. Motivation
Google’s Ngram Viewer reports the relentless growth of Bayesian
networks:
§ Parallel increases in complexity & data.
§ New ways of interactively visualizing both networks & data are needed.

4. Omnigram Explorer
§ Developed by Tim Taylor; open source
– www.tim-taylor.com
– https://github.com/tim-taylor/omnigram
§ Traditional visualization of multiple variables:
– Scatterplots
– Parallel Coordinates: allow tracing samples across many
dimensions

5. Omnigram Explorer
Two main innovations:
§ Visualization takes advantage of human visual system, especially
psychology of pattern and motion perception
§ Interactive controls allow user to manipulate patterns, visualizing not
just static data but active dependencies between variables
In particular, OE enables interactive sensitivity analysis of Bayesian
networks with user-chosen sets of observations (“sensitivity to data”)

6. OE
OE Window
BN based on UCI car+mpg data set

7. OE: Single Node Brushing
§ “1” starts single node brushing
§ Enter observation as a range in one variable; observe consequences
§ Drag range to observe dependencies

8. OE: Multi Node Brushing
§ “2” starts multi node brushing
§ Enter observations as ranges in multiple variables; colors show full & partial
matches
§ Drag ranges to observe dependencies

9. OE: Linked Brushing
§ In multi node brushing, hit “L” with cursor over successive nodes
§ Drag range in either variable, linked variables follow
§ You can reverse link direction (negative dependency); link’s box will
invert

10. OE: Flash Mode
§ “3” starts flash mode
§ Cycles through small (or large) samples, replacing oldest with new
§ Speed and sample size controllable

11. Exploring Dependencies: D-Separation
§ In multi node, fix an observation set (here cylinders & horsepower)
§ Observe which variables respond when varying another (e.g., model year)
§ Partial observations will “leak”; other leakage indicates (some) violation of the
Markov property

12. Exploring Dependencies: Explaining Away
§ In multi node, fix a child of more than one parent; explore the induced
dependency between parents (“explaining away”)
§ Here early model year partially explains high displacement, so Origin=1 (US)
declines

13. Exploring Dependencies: Inference to the Best
Explanation
§ What could explain an unusual event? E.g., high acceleration + good mpg?
§ Read off the means for precursor variables: late model, foreign, 4 cylinders, low
weight, etc.

14. Conclusion
OE is a great interactive tool for either:
§ Exploring a data set, to get a feel for dependencies and
independencies, relations worth exploring
§ Exploring a Bayesian network:
– To get a feel for dependencies and independencies, relations
worth exploring
– Examine d-separation properties
– Performing sensitivity analysis interactively & visually
NB: OE is not (yet) implemented via a BN API, but can be used via
sampling with any BN.

15. References
§ Taylor, T., Dorin, A., & Korb, K. Omnigram Explorer: A Simple
Interactive Tool for the Initial Exploration of Complex Systems.
European Conference on Artificial Life, 2015.
§ Ropero, R. F., Nicholson, A. E., & Korb, K. (2015). Using a New Tool to
Visualize Environmental Data for Bayesian Network Modelling. In
Advances in Artificial Intelligence (pp. 175-184). Springer International
Publishing.