Joshua Romoff
Joshua Romoff
Verified email at mail.mcgill.ca
TitleCited byYear
Hybrid reward architecture for reinforcement learning
H Van Seijen, M Fatemi, J Romoff, R Laroche, T Barnes, J Tsang
Advances in Neural Information Processing Systems, 5392-5402, 2017
832017
Tarmac: Targeted multi-agent communication
A Das, T Gervet, J Romoff, D Batra, D Parikh, M Rabbat, J Pineau
arXiv preprint arXiv:1810.11187, 2018
112018
Multi-advisor reinforcement learning
R Laroche, M Fatemi, J Romoff, H van Seijen
arXiv preprint arXiv:1704.00756, 2017
112017
Randomized value functions via multiplicative normalizing flows
A Touati, H Satija, J Romoff, J Pineau, P Vincent
arXiv preprint arXiv:1806.02315, 2018
92018
Separation of concerns in reinforcement learning
H van Seijen, M Fatemi, J Romoff, R Laroche
arXiv preprint arXiv:1612.05159, 2016
9*2016
Reward estimation for variance reduction in deep reinforcement learning
J Romoff, P Henderson, A Piché, V Francois-Lavet, J Pineau
arXiv preprint arXiv:1805.03359, 2018
42018
Separating value functions across time-scales
J Romoff, P Henderson, A Touati, Y Ollivier, J Pineau, E Brunskill
International Conference on Machine Learning, 5468-5477, 2019
32019
Where did my optimum go?: An empirical analysis of gradient descent optimization in policy gradient methods
P Henderson, J Romoff, J Pineau
arXiv preprint arXiv:1810.02525, 2018
32018
Deep conditional multi-task learning in atari
J Romoff, E Bengio, J Pineau
ICML 48, 2016
32016
Gossip-based Actor-Learner Architectures for Deep Reinforcement Learning
M Assran, J Romoff, N Ballas, J Pineau, M Rabbat
arXiv preprint arXiv:1906.04585, 2019
2019
Scalability of reinforcement learning by separation of concerns
HH Van Seijen, SMF Booshehri, RMH Laroche, JS Romoff
US Patent App. 15/634,811, 2018
2018
Hybrid reward architecture for reinforcement learning
HH Van Seijen, SMF Booshehri, RMH Laroche, JS Romoff
US Patent App. 15/634,914, 2018
2018
About the attractor phenomenon in decomposed reinforcement learning
R Laroche, M Fatemi, J Romoff, H van Seijen
2018
The system can't perform the operation now. Try again later.
Articles 1–13