Commit 2610fca0 authored by Paul Fiterau Brostean's avatar Paul Fiterau Brostean
Browse files

Small corrections.

parent 04b8c7eb
......@@ -61,7 +61,7 @@ We can use Z3GI to learn a model for the observations in `regex_example` as foll
$ python z3gi -m traces -a DFA -f resources\traces\regex_example
```
This produces an output containing:
This produces an output containing a textual description of the learned model:
```
Learned model:
......@@ -81,18 +81,17 @@ q2
```
The model description is split into 3 (or 2) sections.
- the first section indicates acceptance/rejection of each state
- the second section indicates access sequences to the state, in later versions, this section may be ommitted
- the first section describes the transition for the DFA
- the first section indicates acceptance/rejection of each state
- the second section indicates access sequences to each state, in later versions, this section may be ommitted
- the third section describes transitions of the learned model
Learning scalable systems
-----------------------
Z3GI implements several scalable systems such as Sets or Login Systems. What these
systems have in common is that they are configurable in size. A greater size leads
results in a bigger system. Z3GI can be used to learn these systems.
Z3GI implements several scalable systems such as Sets or Login systems. What these
systems have in common is that they are configurable in size. A greater size
results in a larger system. Z3GI can be used to learn these systems.
To give an example, suppose we want to learn a Login system with 2 users
which is structured as a Mealy machine. To learn this system we run:
......@@ -101,6 +100,12 @@ which is structured as a Mealy machine. To learn this system we run:
$ python z3gi -m scalable -a MealyMachine -sc Login -s 2
```
Each scalable system is implemented in many different formalisms. Say we want to learn a
Register Automaton variant of the Login system with 1 user. Then we would have to run:
```
$ python z3gi -m scalable -a RegisterAutomaton -sc Login -s 1
```
Learning .dot models
-----------------------
......@@ -115,14 +120,15 @@ $ python z3gi -m dot -a MealyMachine -f resources\models\biometric.dot
```
For bigger models which are more difficult to test, we may require an external
test algorithm. We provide a Windows binary for the [Yannakakis test algorithm][yan].
You can activate this algorithm using the `-y path_bin` option. By using `-m dotnorst`,
test algorithm. We provide a Windows 64 bit binary for the [Yannakakis test algorithm][yan]
found in the `resources\binaries`. You can activate this algorithm using the
`-y path_bin` option where `path_bin` is a path to the binary. By using `-m dotnorst`,
Z3GI will attempt to learn without using resets.
Our implementation currently only supports Mealy machines, though functionality
Our implementation currently only supports learning .dot Mealy machine models, though functionality
for other formalisms will be added in the future. Note there is currently no
standard .dot format for register automata. Also note that the Yannakakis test
standard .dot format for Register Automata. Also note that the Yannakakis test
algorithm only works on Mealy machines and requires resets.
[yan]: https://gitlab.science.ru.nl/moerman/Yannakakis
......@@ -135,8 +141,8 @@ Z3GI can be used to learn without reset randomly generated Mealy machines with t
property that they are strongly connected (though not necessarily minimal).
Say we want to learn a randomly generated Mealy machine with 2 inputs, 2 outputs
and 3 states. Then we run:
and 3 states. We then run:
```
$ z3gi -m randnorst -a MealyMachine -ni 2 -no 2 -ns 3
```
\ No newline at end of file
```
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment