Papers of the day   All papers

An Explicitly Relational Neural Network Architecture

Comments

May 28 2019 DeepMind

“An Explicitly Relational Neural Network Architecture” - new work from the DeepMind cognition team takes a step towards reconciling #deeplearning and symbolic #AI, with Marta Garnelo, Kyriacos Nikiforou, @ToniCreswell, Christos Kaplanis, and @dgbarrett: https://arxiv.org/abs/1905.10307 https://t.co/hFnbPrhnUa
3 replies, 652 likes


May 28 2019 Murray Shanahan

Very proud to announce new work from the @DeepMindAI cognition team, bridging the gap between #DeepLearning and symbolic #AI, with Marta Garnelo, Kyriacos Nikiforou, @ToniCreswell, Christos Kaplanis, and @dgbarrett - https://arxiv.org/abs/1905.10307
5 replies, 315 likes


Sep 26 2019 Murray Shanahan

A revised version of our paper "An Explicitly Relational Neural Network Architecture" is out, and the PrediNet code and datasets are now open-source. Paper: https://arxiv.org/abs/1905.10307. Code: https://github.com/deepmind/deepmind-research/tree/master/PrediNet. @DeepMindAI
1 replies, 215 likes


Jun 06 2019 Annie Duke

New AI start up, http://robust.ai, merging deep learning with symbolic AI. Will be exciting to watch, especially with @GaryMarcus, @RodneyABrooks, Anthony Jules, @hiskov & @mohamedramer as founders. They are hiring! http://lnkd.in/em9hcpG https://twitter.com/thenextweb/status/1136430072246820867?s=12
1 replies, 20 likes


Nov 30 2019 Murray Shanahan

Indeed. Respect to Josh Tenenbaum & colleagues. My (current) preference is to work with (novel) deep learning architectures (built out of differentiable components), such as https://arxiv.org/abs/1905.10307. But no-one knows what the right approach is yet. Maybe more than one will work
4 replies, 13 likes


Jun 06 2019 Murray Shanahan

"Merging deep learning with symbolic AI" - sounds like a great idea to me. For a step in that direction from my @DeepMindAI team - https://arxiv.org/abs/1905.10307
0 replies, 12 likes


May 28 2019 Murray Shanahan

https://t.co/KR3cDk8N9Y
0 replies, 12 likes


Jun 01 2019 Hacker News

An Explicitly Relational Neural Network Architecture https://arxiv.org/abs/1905.10307
1 replies, 10 likes


Jan 03 2020 Murray Shanahan

@GaryMarcus I think we can build all the parts needed for the right architecture out of differentiable components. This is our current work along these lines, but it's just a beginning: https://arxiv.org/abs/1905.10307
3 replies, 8 likes


May 28 2019 DeepMind

“An Explicitly Relational Neural Network Architecture” - new work from the DeepMind cognition team takes a step towards reconciling #deeplearning and symbolic #AI, with Marta Garnelo, Kyriacos Nikiforou, @ToniCreswell, Christos Kaplanis, and @dgbarrett: https://arxiv.org/abs/1905.10307
0 replies, 7 likes


Sep 27 2019 kaalam.ai

Bridging the gap between deep learning and symbolic AI! Code + paper
0 replies, 6 likes


Nov 30 2019 Juan Valle Lisboa

@GaryMarcus @mpshanahan And that's what Tenenbaum and his inference over programs or game engines is trying to achieve. Really cool work and not DL.
0 replies, 4 likes


Jun 02 2019 Dmitri Sotnikov ⚛

An Explicitly Relational Neural Network Architecture https://arxiv.org/abs/1905.10307
0 replies, 1 likes


May 28 2019 Thomas Scialom

A must read from deepmind!
0 replies, 1 likes


Content