aka learning-to-learn

Deep meta-reinforcement learning

Flexible, data-efficient learning naturally requires the operation of prior biases. In general terms, such biases can derive from two sources; they can either be engineered into the learning system (as, for example, in convolutional networks), or they can themselves be acquired through learning. The second case has been explored in the machine learning literature under the rubric of meta-learning (Schmidhuber et al., 1996; Thrun and Pratt, 1998).

In one standard setup, the learning agent is confronted with a series of tasks that differ from one another but also share some underlying set of regularities. Meta-learning is then defined as an effect whereby the agent improves its performance in each new task more rapidly, on average, than in past tasks (Thrun and Pratt, 1998). At an architectural level, meta-learning has generally been conceptualized as involving two learning systems: one lower-level system that learns relatively quickly, and which is primarily responsible for adapting to each new task; and a slower higher-level system that works across tasks to tune and improve the lower-level system

(from here)

https://2017.icml.cc/Conferences/2017/Schedule?showEvent=495