Artificial Intelligence


Deep Meta-Learning and Dynamic Runtime Exploitation of Knowledge Sources for Traffic Control

Authors: Sandra Ottl

In the field of machine learning and artificial intelligence, meta-learning describes how previous learning experiences can be used to increase the performance on a new task. For this purpose, it can be investigated how prior (similar) tasks have been approached and improved, and knowledge can be obtained about achieving the same goal for the new task. This paper outlines the basic meta-learning process which consists of learning meta-models from meta-data of tasks, algorithms and how these algorithms perform on the respective tasks. Further, a focus is set on how this approach can be applied and is already used in the context of deep learning. Here, meta-learning is concerned with the respective machine learning models themselves, for example how their parameters are initialised or adapted during training. Also, meta-learning is assessed from the viewpoint of Organic Computing (OC) where finding effective learning techniques that are able to handle sparse and unseen data is of importance. An alternative perspective on meta-learning coming from this domain that focuses on how an OC system can improve its behaviour with the help of external knowledge sources, is highlighted. To bridge the gap between those two perspectives, a model is proposed that integrates a deep, meta-learned traffic flow predictor into an organic traffic control (OTC) system that dynamically exploits knowledge sources during runtime.

Comments: 7 Pages.

Download: PDF

Submission history

[v1] 2019-03-05 09:07:04

Unique-IP document downloads: 24 times is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. will not be responsible for any consequences of actions that result from any form of use of any documents on this website.

Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.

comments powered by Disqus