Papers of the day   All papers

Extending Machine Language Models toward Human-Level Language Understanding

Comments

DeepMind: What does it mean to understand language? We argue that human-like understanding requires complementary memory systems and rich representations of situations. A roadmap for extending ML models toward human-level language understanding: https://arxiv.org/abs/1912.05877 https://t.co/eKBOekfmgj

10 replies, 777 likes


Felix Hill: It is commonly said that models like BERT or GPT-2 don't really 'understand', but what does it actually mean to understand language? We try to answer this via a roadmap for human-like understanding of language in machines. https://arxiv.org/abs/1912.05877

6 replies, 392 likes


Anirudh Goyal: "Extending Machine Language Models toward Human-Level Language Understanding" by Jay McClelland, with @jasonbaldridge, @FelixHill84, Maja Rudolph, Hinrich Schutze. https://arxiv.org/abs/1912.05877 is interesting read (Even if you aren't interested in language)

2 replies, 42 likes


Jason Baldridge: New paper on extending ML models toward human-level language understanding! It's a joint effort with Jay McClelland, @FelixHill84, Maja Rudolph, and Hinrich Schütze that integrates our diverse perspectives on cognition, grounding, modeling and language. https://arxiv.org/abs/1912.05877 https://t.co/3BzJVL7l6W

3 replies, 36 likes


Carlos E. Perez: It's strange to me that Deep Learning researchers tie language capabilities back to the human brain when it's known that all of life is based on language interpretation. https://arxiv.org/abs/1912.05877

2 replies, 18 likes


Jason Baldridge: I agree! This is the key principle in my current research. See: - Baldridge et al (2018) Points, Paths, and Playscapes (https://www.aclweb.org/anthology/W18-1406/) - McClelland et al (2020) Extending Machine Language Models toward Human-Level Language Understanding (https://arxiv.org/pdf/1912.05877.pdf)

1 replies, 12 likes


Ronen Tamari: @yoavgo Really identify with this. Also- increasingly, people are talking about the need for situated/grounded language understanding, world models, etc (e.g. https://arxiv.org/abs/1912.05877, https://docs.google.com/presentation/d/1DWakzn8cQfzMC7dOaLNWkFBWyKwmSD6Wc7F-vy_MKLw/edit#slide=id.g59d392dcb8_0_584)

2 replies, 4 likes


arXiv CS-CL: Extending Machine Language Models toward Human-Level Language Understanding http://arxiv.org/abs/1912.05877

0 replies, 3 likes


Project AGI: “Extending Machine Language Models toward Human-Level Language Understanding” Very different application areas. https://arxiv.org/abs/1912.05877

0 replies, 2 likes


Andreas Hanselowski: Interesting to see that big players in @AI are moving to grounded language learning: @Montreal_AI https://silo.ai/research-club-babyai/ @facebookai https://www.youtube.com/watch?v=gnjzTaHmnug @DeepMindAI https://deepmind.com/research/publications/understanding-grounded-language-learning-agents https://arxiv.org/abs/1912.05877 https://slideslive.com/38921905/visually-grounded-interaction-and-language-1

0 replies, 1 likes


Will Rice: Do we really understand language?

0 replies, 1 likes


Julian Pani: Good paper to get some perspective on the state of NLP as we go into the 2020's

0 replies, 1 likes


Adam Santoro: Had the pleasure of reading a draft of this, and I cannot recommend it enough. Please read it if you're at all interested in the role of language in intelligence

0 replies, 1 likes


Ton Ngo: Episodic memory! @JimSpohrer

0 replies, 1 likes


Content

Found on Dec 16 2019 at https://arxiv.org/pdf/1912.05877.pdf

PDF content of a computer science paper: Extending Machine Language Models toward Human-Level Language Understanding