# ML Wiki

## Workflow Nets

By workflow nets here we refer to petri-net-based workflows.

So, a workflow net is a special type of a petri net that is suitable for expressing workflows

In a workflow net:

• there's a clear start:
• the unique dedicated input place $i$ s.t.
• $\bullet i = \varnothing$
• i.e. no transition can put a token to $i$
• there's a clear end:
• the unique dedicated output place $o$ s.t.
• $o \bullet = \varnothing$
• i.e. no transitions should consume tokens from $o$
• every other transition and place are on the path from $i$ to $o$

Markings:

• initial marking $M_0$:
• $M_0(i) = 1, \forall p \ne i: M_0(p) = 0$

### Examples

Not workflow nets

• (1): no start, no end
• (2): $b$ is not connected to the end
• (3): $d$ is also not on the path from $i$ to $o$

Workflow net:

• all workflow net properties are satisfied

## Soundness

Main Article: Workflow Soundness

There are situations that we want to avoid:

• unboundness

For that we define the following properties

• Option To Complete
• Proper Termination

## Workflow Patterns

Main Article: Workflow Nets/Workflow Patterns

## Sources

Machine Learning Bookcamp: Learn machine learning by doing projects. Get 40% off with code "grigorevpc".