Answers are grounded in this paper and cite their sources.
In this paper we review recent results on the adaptive behavior attained with fixed-weight recurrent neural networks (meta-learning). We argue that such behavior is a natural consequence of prior training.
No takes yet. Share an insight, caveat, or question.
Prokhorov et al. (2003) studied this question.