Neuro-symbolic Models for NLP (6.884)

Jacob Andreas (jda@mit.edu)

Class will meet on Fridays from 11:30 AM to 1:30 PM ET. We'll use this zoom link. Discussion will take place on Piazza.

Office hours: Thursdays 2pm–4pm (by appointment).

1/3 in-class paper presentation, 1/3 participation and reading responses, 1/3 final project.

Readings and presentation signups on this spreadsheet.

Intro		[slides:pdf]
Symbols & systematicity	Fodor & Pylyshyn 1998 Lake & Baroni 2018	[slides:pdf]
Other kinds of generalization	Hu et al. 2020 Sinha et al. 2019	Kamoya Ikhofua and Greta Tuckute [slides:pdf]
Empirical successes	Furrer et al. 2020 Rumelhart & McClelland 1985	Simon Alford and Mahi Elango [slides:pdf]
Connectionist symbol processing	Smolensky 1990 Plate 1990	Joey Velez-Ginorio, Tiwalayo Eisape and Pedro Colon-Hernandez [slides:pdf]
Discrete representations	Weiss et al. 2018 Dalvi et al. 2018	Evan Hernandez and Geeticka Chauhan [slides:pdf]
Modular representations	McCoy et al. 2019 Andreas 2019	Rami Manna, Hope Kean and Ekin Akyürek [slides:pdf]
Modular computation	Geiger et al. 2020 Partee 1984	Carina Kauf, Matthew Huggins, Yixuan Pei and Hang Jiang [slides:pdf]
Tree-shaped models	Socher et al. 2013 Dyer et al. 2016 Andreas et al. 2016	Ian Palmer, Shinjini Ghosh and Lara Rakocevic [slides:pdf]
Structured losses	Oh et al. 2017 Mu et al. 2020	Belén Saldías [slides-a:pdf] [slides-b:pdf]
Meta-learning	Lake 2019 McCoy et al. 2020	Ferran Alet, Abby Bertics and Joe O'Connor [slides:pdf]