Differences

This shows you the differences between two versions of the page.

--- nlp:learning_to_communicate [2023/11/03 23:06] – [Papers] jmflanig
+++ nlp:learning_to_communicate [2025/04/30 20:32] (current) – [Papers] jmflanig
@@ Line 1: / Line 1: @@
 ====== Learning to Communicate ======
 There are some papers on artifical agents learning to communicate to play a game, etc.  They either learn a language from scratch (an emergent language) or use reinforcement learning to improve an existing model of language.
+===== Overviews =====
+  * Multi-agent communication
+    * [[https://link.springer.com/article/10.1007/s10462-021-09996-w|2020 - Multi-agent deep reinforcement learning: a survey]]
 ===== Papers =====
-  * [[https://proceedings.neurips.cc/paper_files/paper/2017/file/70222949cc0db89ab32c9969754d4758-Paper.pdf|Havrylov & Titov 2017 - Emergence of Language with Multi-agent Games: Learning to Communicate with Sequences of Symbols]]
+  * **Multi-agent**
-  * [[https://arxiv.org/pdf/2006.02419.pdf|2020 - Emergent Multi-Agent Communication in the Deep Learning Era]]
+    * [[https://proceedings.neurips.cc/paper_files/paper/2017/file/70222949cc0db89ab32c9969754d4758-Paper.pdf|Havrylov & Titov 2017 - Emergence of Language with Multi-agent Games: Learning to Communicate with Sequences of Symbols]]
+    * [[https://aclanthology.org/P19-1509.pdf|2019 - Word-order biases in deep-agent emergent communication]]
+    * [[https://arxiv.org/pdf/2006.02419.pdf|2020 - Emergent Multi-Agent Communication in the Deep Learning Era]]
+    * [[https://arxiv.org/pdf/2012.08630.pdf|2020 - Open Problems in Cooperative AI]]
+  * **Emergence of Grammar**
+    * [[https://arxiv.org/pdf/2503.01635?|Wechsler et al 2025 - The Emergence of Grammar through Reinforcement Learning]]
 ===== Related Pages =====