OpenPoll - small polls with open-ended answers



Are you using Adobe Lightroom?

The new FREE plugin Mykeyworder for Lr can SIGNIFICANTLY speed up your keywording process.

Read my article
How to Gain Influence by Taking Initiative in Day-to-Day Situations
Become more assertive and influential
A clap will support this site

Read my article
5 Techniques for Forming, Changing, and Breaking Habits
Being mentally strong is not enough!
A clap will support this site

Read my article
Why the old text editor 'Emacs' is still in my toolbox
First version is from 1985 — software rarely survives that long!
A clap will support this site



DeepRL NIPS 2016 Panel Discussion Questions

What are the pros and cons of the different RL benchmark suites (OpenAI Gym/Universe, DeepMind Lab, Atari, etc.)?
How can a newbie get started in deep RL?
What factor is limiting progress in deep RL?
Can small academic labs keep up with industry?
What is the biggest idea that we are currently missing?
When should one use value-based learning (DQN) and when should one use policy gradient based methods (A3C, TRPO)?
Why do we believe robotics with RL is solvable if guided policy search is inefficient and cannot generalise across objects and work over long horizons? How do we go about in trying to solve it?
How can we provide video game ROMs legally? Today, many researchers download and use illegal ROMs e.g. Atari.
What are the most promising approaches to emergent hierarchy in RL systems?
Should we be aiming for Human Level AI or something else altogether?
A lot of recent methods seem to be learning additional environment parameters. Is this a step towards model based RL?
What are the implications of progress in guided cost learning for AI safety?
What are the biggest obstacles for model-based (or hybrid) RL with deep NNs? Why so much focus on model-free RL, even in domains like robotics, where sample efficiency is of immense importance?
How might we connect deep RL to universal AI work from Hutter/Solomonoff?
Other 

        
Terms of service · Privacy policy
Please donate to keep OpenPoll running.
A donation of at least 2$ will remove ads and the Donate button for this poll.
Amount:  USD

Please support OpenPoll by signing up to Presearch - you will protect your privacy and get 25 PRE tokens
currently worth around 2.50$, remove ads on this poll (worth 2$), and get one month ad-free OpenPolls (worth 5$).

Click here to claim after signing up via the link below.

presearch