Denis Yarats

Denis's broad research interests include reinforcement learning, optimal control, and robotics. His specific focus is on representation learning within RL that can enable sample-efficient algorithms with applications in robotics.

Denis' Work

Denis's Publications

May 03, 2020

RESEARCH

COMPUTER VISION

Quasi-Hyperbolic Momentum and Adam for Deep Learning

Momentum-based acceleration of stochastic gradient descent (SGD) is widely used in deep learning. We propose the quasi-hyperbolic momentum algorithm (QHM) as an extremely simple alteration of momentum SGD, averaging a plain SGD step with a…

Jerry Ma, Denis Yarats,

May 03, 2020

May 03, 2020

RESEARCH

NLP

Hierarchical Text Generation and Planning for Strategic Dialogue

End-to-end models for goal-orientated dialogue are challenging to train, because linguistic and strategic aspects are entangled in latent state vectors. We introduce an approach to learning representations of messages in dialogues by maximizing…

Denis Yarats, Mike Lewis,

May 03, 2020

May 03, 2020

RESEARCH

SPEECH & AUDIO

Convolutional Sequence to Sequence Learning

The prevalent approach to sequence to sequence learning maps an input sequence to a variable length output sequence via recurrent neural networks. We introduce an architecture based entirely on convolutional neural networks.1 Compared to…

Jonas Gehring, Michael Auli, David Grangier, Denis Yarats, Yann N. Dauphin,

May 03, 2020