Devendra Singh Chaplot, Kanthashree Mysore Sathyendra, Rama Kumar Pasumarthi, Dheeraj Rajagopal, Ruslan Salakhutdinov
Carnegie Mellon University
{chaplot,ksathyen,rpasumar,dheeraj,rsalakhu}@cs.cmu.edu
Multitask Learning: The agent is evaluated on unseen maps with instructions in the train set. Unseen maps comprise of unseen combination of objects placed at randomized locations. This scenario tests that the agent can execute multiple instructions or tasks in unseen maps.
Zero-shot Learning: The agent is evaluated on unseen test instructions. This scenario tests whether the agent can generalize to new combinations of attribute-object pairs which are not seen during the training. The maps in this scenario are also unseen.
Full article: https://arxiv.org/pdf/1706.07230.pdf
Code, Environment, Pre-trained models: https://github.com/devendrachaplot/DeepRL-Grounding
Media: MIT TechReview, Inverse