COMS W4705: Homeworks

 Gilbert Street
[Main] | [General Information] | [Problem Sets]


  • October 9th: Homework 2 is released here. The analytical problems are due by 5pm on October 23rd; the programming assignment is due by 5pm on October 30th. Files for the programming assignment: hw2.tar.gz.

  • September 11th: Homework 1 is released here. The analytical problems are due September 25th; the programming assignment is due October 2nd (see the homework for full details). Files for the programming assignment: count_freqs.py, eval_ne_tagger.py, ner_train.dat, ner_dev.dat, ner_dev.key.

    An important update: please read this note on a modified version of the Viterbi algorithm, using log probabilities. You should implement this modified version.

  • November 12th: Homework 3 analytical problems are released here. The homework is due on November 27th.

  • November 27th: Homework 4 is released here. The homework is due on December 11th. The github link for the code for homework 4 is here. Sample code for a neural network part-of-speech tagger is here.

    Submission Instructions

  • Analytical part: problems must be uploaded to Courseworks. For handwritten homeworks please scan your homework to create a pdf file and then upload the file.
  • Programming assignments: Please see the instructions on Piazza for how to submit.
  • Update on late policy: we will give students 5 "free" days that can be used as they wish across the 4 problem sets. Specifically, we will not penalize the first 5 late days that a student incurs on problem sets. After that, the penalties posted on the problem sets will apply (e.g., 5 points per day late on the first problem set). The final (0 point) deadline will still apply.

    Programing Assignments Policy and Guidelines

    1. Document your code! Undocumented code will result in lower scores.
    2. Write a brief report describing results of experiments, any observations you made, design choices and instructions on how to build (if necessary) and run your implementation (command line arguments, whether data is fed to your program on stdin or from a file, etc.). The report is part of your solution and will be scored. It can be in plain text or PDF.
    3. Make sure your program implements any specific functionality we ask for (input/output format etc.).
    4. Efficiency of your implementation matters only when we ask for it (your algorithms should have desired performance and space requirements).
    5. You should be able to solve all problems using pre-installed standard libraries. Do not use any NLP or machine learning libraries. If you choose to use third-party libraries or modules (e.g numeric computing frameworks such as numpy), make sure they are installed on CLIC. When in doubt if it is okay to use third-party code ask the TAs.
    6. Please separate your report into sections. For example, for each problem describe
      Part1: how to run your code step by step (make sure your code can run on CLIC).
      Part2: performance for your algorithm (including precision, recall, and F-score).
      Part3: observations and comments about your experimental results.
      Part4: any additional information that is requested in the problem.

      Group Work and Academic Honesty Policy

      All problems must be solved individually. You may discuss the problems with other students, but you have to do the write-up and implementation yourself. We will check homework assignments for duplicates. Violations will result in a grade of zero and further steps may be taken in accordance with the CS department's academic honesty policy.