Kevin Roice

Reinforcement Learning @AmiiThinks. Previously fellow @rlai_lab, @cambridge_uni, Physics & CS @durham_uni

[Martin Bauer @martinmbauer]

Physics and computer science is almost a parent-child relationship. That's why many computer scientists have it in their DNA, whether they realise it or not

[Khurram Javed @kjaved_]

World models are all the rage these days, so it's worth reiterating a few points that are largely correct.

1. Yes, our agents need models. The primary use of these models is planning. Planning can be done in real time, to improve an immediate decision, or in the background when not much is going on, to improve future decisions.

2. Learning models that predict the next sensory percept, such as pixels, is insufficient. The models should predict agent state; agent state is a summary of the past observations.

3. Learning one-step models is insufficient. Models should be conditioned on sequences of actions (e.g., option models). Finding what sequences of actions they should be conditioned on is an unsolved problem.

[Simon Cornish @SimonLCornish]

Our paper on entanglement of molecules was published in Nature today!
@DurhamQLM @DurhamPhysics
https://t.co/xuPSQZJePy

[Journal of Machine Learning Research @JmlrOrg]

'Empirical Design in Reinforcement Learning', by Andrew Patterson, Samuel Neumann, Martha White, Adam White.

https://t.co/HE6kB5jQLH

#reinforcement #experiments #benchmark

[Marlos C. Machado @MarlosCMachado]

RLC will be held at the Univ. of Alberta, Edmonton, in 2025. I'm happy to say that we now have the conference's website out: https://t.co/ZjpvWi5jyV

We'll continue to update it, and the CFP will be out soon, but the relevant dates are already there.

@RL_Conference @UAlberta

[Joseph Suarez 🐡 @jsuarez]

Why I think most RL papers are probably wrong in 1 graph. This is a 100 experiment hyperparam sweep for 200M steps each. Didn't find anything useful until >10B samples. Showing anything at all about an algorithm requires extensive sweeps that are rarely done!

[Richard Sutton @RichardSSutton]

A year later and our work on Loss of Plasticity is finally published, in Nature no less! The Nature version is totally rewritten and has many new results:
https://t.co/QImypXpqQl
Congratulations to the authors:
@s_dohare
@JFernandoHG
@LanceLan3
@rahman_parash
@rupammahmood

[Richard Sutton @RichardSSutton]

Yesterday there was a completely-student-organized summit of the RLAI (Reinforcement Learning and Artificial Intelligence) research group at the University of Alberta, held at the lovely Amii headquarters. Nice folks and diverse new ideas!

[Kevin Roice @KevinTheRoice]

@pcastr An old one is the PinBall environment:
https://t.co/heF8VQsChg

[Esra'a Saleh @TheEsraaSaleh]

Would AI researchers (students, profs and alumni) of Canadian universities be interested in having one unified Slack that we collectively own?

I've seen a lot of scientific and career value coming from the simple fact that we can communicate and share experience.

[The Cultural Tutor @culturaltutor]

A short history of the world in 13 maps & infographics:

1. Everybody alive today compared to everybody who has ever lived

[serif @serifhello]

Wait a second, these are not real error bars … the author literally just put the letter “T” above the bar graphs 😭

[Amazing Maps @amazingmap]

Sunshine duration in hours per year. USA and Europe

[Kevin Roice @KevinTheRoice]

@karpathy @GoogleMagenta 👀

[Patrick Mineault @patrickmineault]

One of the wildest recent findings in matching AI vision to human vision is the surprising importance of adversarial robustness. 1/

[The New York Times @nytimes]

The front page of The New York Times for May 24, 2020

[Domenico @AvatarDomy]

A marine biologist films this octopus changing colors while dreaming and it’s spectacular

[OREO Cookie @Oreo]

Don't have a bookmark? Try using OREO cookies and milk instead:

[Aero @ActualAero]

Ash Ketchum has been a part of millions of lives & while many stop watching the Pokémon anime Ash continued to be a role model trainer for new generations. He's lost every Pokémon League to teach children it's ok to lose & today he finally won. What a day to be a Pokémon fan :)