Opportunistic Knowledge Adaption in Self-Learning Systems
Simon Reichhuber, Sven Tomforde · 2020
In the context of Autonomous Learning, the question arises how an online learning system adapts its knowledge according to a changing environment, i.e. arrival of new classes or changing noise functions, to maintain a robust level of performance. As a solution, we suggest an architectural design inspired by a variant of the Observer/Controller framework. We present a scenario, in which the presented architecture is assumed to improve the performance, because the system is aware of currently available knowledge and can opportunistically exploit this knowledge.