Schedule

Plan Code/Data Screencasts/Readings Assignments
Mar 30
  1. Course overview
    [slides]
  1. Install the latest iPython [here are our tips on this]
  2. Install Numpy, Scipy, and matplotlib. (Mac users: we suggest the Scipy Superpack)
  3. Install scikit-learn
  1. Tou and Zelle 1997
  2. Ferrucci et al. 2010
  3. Mitchell 2004
  4. Podcast: The challenge and promise of artificial intelligence
  5. Levesque 2013
  1. HW 1 [due Apr 15]: From the Distributed word representations codelab exercises, complete any 12 distinct exercises in the “Straightforward” and/or “Challenging” sections.
Apr 1
  1. Distributed word representations
  1. Codelab in HTML
  2. Github codebase
  3. Data distribution
  4. Pre-computed GloVe vectors for the IMDB data
  5. Insights from the first in-class bake-off
  6. In-class bake-offs
  7. Word entailment bake-off materials (also in the Github repository)
  1. Turney and Pantel 2010
  2. Screencast: Overview [slides]
  3. Screencast: Vector comparison [slides]
  4. Screencast: Reweighting [slides]
  5. Screencast: Dimensionality reduction [slides]
  6. Talk: t-SNE (van der Maaten)
  7. Screencast: Word-sense disambiguation [slides]
  8. Talk: GloVe (Pennington)
Apr 6
Apr 8
Apr 13
Apr 15
  1. Workshop 1: Project planning
    [slides]
  1. Domingos 2012
  2. Resnik and Lin 2010
  3. Smith 2011, Appendix B
  1. HW 2 on relation extraction [due Apr 27]
Apr 20
  1. Relation extraction [slides]
  1. Jurafsky and Martin 2009, §22
  2. Snow et al. 2005
  3. Mintz et al. 2009
  4. Banko et al. 2007
  5. Fader et al. 2011
  6. Yao et al. 2012
Apr 22
Apr 27
  1. Semantic parsing
    [slides]
  1. SippyCup codebase
  2. SippyCup unit 0
  1. Screencast: Core concepts for semantic parsing [slides]
  2. Screencast: Semantic parsing models [slides]
  3. Liang and Potts 2015
  4. Zettlemoyer & Collins 2005
  5. Liang et al. 2013
  6. Talk: Learning dependency-based compositional semantics (Percy Liang)
  7. Kwiatkowski et al. 2013
  1. HW 3 consists of seven problems (all required). It is distributed in three content-identical formats (html, py, ipynb) as part of the SippyCup codebase. You're encouraged to turn your work in as iPython HTML output or as a Python script [due May 11].
  2. Lit review due May 6
Apr 29
  1. SippyCup unit 1
  2. The math (imminent alien invasion) bake-off (drag the files into SippyCup)
May 4
  1. SippyCup unit 2
  2. The dateparse bake-off (drag the files into SippyCup)
May 6
  1. SippyCup unit 3
May 11
  1. Text to 3D scene generation (Will Monroe) [slides]
May 13
  1. Workshop 2: Writing up and presenting your work
  1. Stuart Schieber on reporting research results
  2. David Goss on math style
  1. HW 4 consists of four problems (all required), which appear at the end of the natural language inference codelab. This is distributed in three content-identical formats (nli.html, nli.py, nli.ipynb) in the CS224u codebase. You're encouraged to turn your work in as iPython HTML output or as a Python script [due May 20].
  2. Project milestone due May 25
May 18
  1. SEMPRE: Semantic Parsing with Execution (Percy Liang)
  2. SEMPRE grammar constructed during class
  1. SEMPRE homepage
May 20
  1. Neural networks for natural language understanding (Sam Bowman)
  1. Natural language inference codelab in HTML
  2. Github codebase
  1. Bowman et al. 2015
May 25
  1. Memorial Day (no class)
May 27
  1. Dialogue
  1. Final project due Jun 10, 3:15 pm
Jun 1
  1. Student presentations
Jun 3