Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
1–10 of 58 posts
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#2This sounds interesting. I will like someone from the field of genetic programming on how this works and how it differs from current DL approaches.
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#3Those are really old results. They should compare to this one: https://arxiv.org/pdf/1511.06581.pdf
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#4Those are really old results. They should compare to this one: https://arxiv.org/pdf/1511.06581.pdf
this is old too
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#5Those are really old results. They should compare to this one: https://arxiv.org/pdf/1511.06581.pdf
How can the results be old when the paper is from 2017?
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#6The convenient thing about Atari games is that there is usually a numerical score that can be used as input for the fitness function.
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#7Those are really old results. They should compare to this one: https://arxiv.org/pdf/1511.06581.pdf
How can the results be old when the paper is from 2017?
They are comparing their genetic programming results with a deep learning paper published in 2015. [1]
[1] https://www.nature.com/nature/journal/v518/n7540/abs/nature1...
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#8This sounds interesting. I will like someone from the field of genetic programming on how this works and how it differs from current DL approaches.
Kelly's approach involves evolving teams of programs.
His basic strategy is to have a scalable problem decomposition strategy.
So programs that process pixels and the teaming of those programs are grouped together. The groupings (teams) themselves are co-evolved with the programs, simultaneously.
This enables niching and specialization behavior.
This builds on earlier work on 'symbiotic bid-based genetic programming' from other people at Dalhousie, the same university Kelly is at.
The innovation of this paper is that teams can reference other teams.
This allows for the creation of hierarchical teams. (There are rules to prevent cycles and other edge cases.)
Everyone commenting here is probably going to just look at numerical game score and ignore the fact that the runtime performance of Kelly's tangled program graphs. They are 1000 times smaller than a deep neural network. That matters for things like running on mobile/embedded devices.
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#9One of the huge benefits of GPs over NNs is the ease of reverse engineering a GP tree compared to NN models. Its not effortless however. Its just not mathematically complex like NNs i.e. a programmer who isn't a mathematician can analyze GPs with a lot of patience
EDIT: I have found GPs to be relatively slow-to-very-slow. But very likely that is because of the lack of interest and development compared to NNs
Re: Multi-Task Learning in Atari Video Games with Emergent Tangled Program Graphs
#10Earlier quoted context omitted.
How can the results be old when the paper is from 2017?
They are comparing their genetic programming results with a deep learning paper published in 2015. [1] [1] https://www.nature.com/nature/journal/v518/n7540/abs/nature1...
What a time to live in, when papers from 2015 are "really old" in 2017.