Andreas Vlachos, Gerasimos Lampouras
{a.vlachos,g.lampouras}@sheffield.ac.uk
Department of Computer Science
University of Sheffield
Sebastian Riedel
s.riedel@ucl.ac.uk
Department of Computer Science
University College London
Imitation learning: an advanced behavior whereby an individual observes and replicates another's behavior
Autonomous helicopter flight
(Coates et al., 2008)
And more: outdoor navigation (Silver et al., 2008), Super-Mario (Ross et al., 2011), autonomous driving (Zhang and Cho, 2017)...
Dynamic oracles for parsing
(Goldberg and Nivre, 2012 )
Incremental coreference resolution
(Clark and Manning, 2015)
Recurrent Neural Network training
(Ranzato et al., 2016)
Search-based structured prediction (Daumé III et al., 2009)
Meta-learning: better model (≈policy) by generating better training data from demonstrations.
Yes: we assume gold standard
output for training
But: we train a classifier to predict
actions constructing the output.
Actions not in gold;
IL is rather semi-supervised
In NLP we train classifiers to imitate experts in many tasks:
Imitation learning has been used to improve accuracy in all the above with SOTA results!
Imitation learning algorithms:
Interpretations and connections
Applications:
Practical advice
Understand how IL works via unified algorithmic presentations
Clarify its connections to other learning frameworks
Know representative NLP applications
Recognize when and how to apply IL