Source Code - A Robust Decision-Making Framework Based on Collaborative Agents

This research item contains an archive with source code and data for reproducing the results reported in the artcile "A Robust Decision-Making Framework Based on Collaborative Agents".

SOFTWARE REQUIREMENTS

To run the attached Python code you need the following libraries to be installed: osbrain, numpy, neat, gzip, pickle, skfuzzy, skimage, PIL, scipy.

GUIDELINES

To produce results, go though the following steps:

1. generate the feed-forward artificial neural networks (ffANN) and genetic fuzzy system (GFS) with the file "Train";

2. copy the produced files relative to the ffANN in file "Net" and the GFS files to the "FS" file, both inside the file "MAS_execution".

3. select the number of agents in the main file of execution (run_process);

4. check the file "AgentX" to modify the source of the information.

5. And finally, you can execute the system running the "run_process" file, each agent will execute as a subprocess.

Additional information:

Folder "Train":

Inside this folter there are two intelligent methods, GFS and NEAT.

*In GFS, you can train modifing file "execution". In this file you can edit the database and the number of agents.

*In NEAT, You can train with "training_Neat". The training parameters configuration per data set are adjustables on files "config-CX"

Both folders has data from four datasets of example.

Folder MAS_execution":

In the folder "Data" and "Test" the system is going to save information of the result system.

The files "definitions", "Features", "GFS_Agents" are libraries that the MAS use.