Modeling Adaptive Learning Agents for Domain Knowledge Transfer
The implementation of intelligent agents in industrial applications is often prevented by the high cost of adopting such a system to a particular problem domain. This paper states the thesis that when learning agents are applied to work environments that require domain-specific experience, the agent...
Saved in:
| Published in: | 2019 ACM/IEEE 22nd International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C) pp. 660 - 665 |
|---|---|
| Main Author: | |
| Format: | Conference Proceeding |
| Language: | English |
| Published: |
IEEE
01.09.2019
|
| Subjects: | |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The implementation of intelligent agents in industrial applications is often prevented by the high cost of adopting such a system to a particular problem domain. This paper states the thesis that when learning agents are applied to work environments that require domain-specific experience, the agent benefits if it can be further adapted by a supervising domain expert. Closely interacting with the agent, a domain expert should be able to understand its decisions and update the underlying knowledge base as needed. The result would be an agent with individualized knowledge that comes in part from the domain experts. The model of such an adaptive learning agent must take into account the problem domain, the design of the learning agent and the perception of the domain user. Therefore, already in the modeling phase, more attention must be paid to make the learning element of the agent adaptable by an operator. Domain modeling and meta-modeling methods could help to make inner processes of the agent more accessible. In addition, the knowledge gained should be made reusable for future agents in similar environments. To begin with, the existing methods for modeling agent systems and the underlying concepts will be evaluated, based on the requirements for different industrial scenarios. The methods are then compiled into a framework that allows for the description and modeling of such systems in terms of adaptability to a problem domain. Where necessary, new methods or tools will be introduced to close the gap between inconsistent modeling artifacts. The framework shall then be used to build learning agents for real-life scenarios and observe their application in a case study. The results will be used to assess the quality of the adapted knowledge base and compare it to a manual knowledge modeling process. |
|---|---|
| DOI: | 10.1109/MODELS-C.2019.00101 |