Timezone: »

QMDP-Net: Deep Learning for Planning under Partial Observability
Peter Karkus · David Hsu · Wee Sun Lee

Tue Dec 05 06:30 PM -- 10:30 PM (PST) @ Pacific Ballroom #113 #None

This paper introduces the QMDP-net, a neural network architecture for planning under partial observability. The QMDP-net combines the strengths of model-free learning and model-based planning. It is a recurrent policy network, but it represents a policy for a parameterized set of tasks by connecting a model with a planning algorithm that solves the model, thus embedding the solution structure of planning in a network learning architecture. The QMDP-net is fully differentiable and allows for end-to-end training. We train a QMDP-net on different tasks so that it can generalize to new ones in the parameterized task set and “transfer” to other similar tasks beyond the set. In preliminary experiments, QMDP-net showed strong performance on several robotic tasks in simulation. Interestingly, while QMDP-net encodes the QMDP algorithm, it sometimes outperforms the QMDP algorithm in the experiments, as a result of end-to-end learning.

Author Information

Peter Karkus (NUS)
David Hsu (National University of Singapore)
Wee Sun Lee (National University of Singapore)

More from the Same Authors