Published December 20, 2023 | Version v1
Dataset Open

20231220-MTOA: Multitasking agent populations can achieve equitable task exploration to the detriment of their agents' average and maximum accuracy.

  • 1. INRIA


This archive contains the results of a multi-agent simulation experiment [1] carried out with Lazy lavender [2] environment.

Experiment Label: 20231220-MTOA

Experiment design: Agents improve task equitability by favoring the reproduction of agents with rare and conflicting knowledge.

Experiment setting: Each agent is initially trained on all tasks. The agents then carry all tasks. When they disagree the following take place:

(a) The agent with the lower score will adapt its knowledge accordingly. If its memory limit is attained, the agent will try to forget knowledge.

(b) The agent with the highest score will decide for both agents. If this agent's decision is correct, the agent will receive the points corresponding to both agents.

Every 20000 games, 9 new agents are born and 9 agents are removed. Each new agent will be trained with examples provided by two parents.The first parent is selected either randomly, or based on their collected points/success rate (low success rate, high success rate, low points, high points).

The second parent is selected based on its success rate with respect to the first parent (parent 2 is the the agent that agrees more often with parent 1).Agents undertake 3 tasks having a limited memory, enough for learning 1/3 tasks accurately (4 and 12 classes respectively)

Hypotheses: Favoring the reproduction of agents with the lowest success rate (agents that agree less with their peers) will allow agent populations to equally explore all tasks. Hence, agent populations will become equally accurate on all tasks and thus improve their efficiency.

Detailed information can be found in index.html or notebook.ipynb.



Files (2.5 GB)

Name Size Download all
2.5 GB Preview Download

Additional details