CSCI-737: Pattern Recognition (Spring 2018)

  RIT Department of Computer Science

CSCI-737-01 Pattern Recognition (Spring 2018)

Instructor: Prof. Richard Zanibbi (web page; rxzvcs at rit dot edu )
Office Hrs: W 1:30-2:30pm, F 12:30-2:30pm
Lectures: MWF 2:30-3:25pm, GOL-3445

News --- Schedule --- Syllabus --- Resources --- MyCourses


Week 15

  • Project 3 is due Thursday at 11:59pm.
  • Please take a moment to fill out the course evaluation - I would greatly appreciate knowing what you thought worked and what could improve so that the course can be improved for future students.
  • We will look at the bonus for Project 2, discuss Project 3, and have a brief review during class on Monday.
  • Prof. Zanibbi will not be holding office hours during the exam period, but will be available by email or by appointment.
  • Best of luck with your final projects and exams!

Week 14

  • !! Final Extension: Project 3 is due next Thursday (May 3rd).
    • Note: There will be no fifth assignment; the weight of the 3rd project will be 20% of the final grade. Just for information, I will post what I had imagined for the fifth assignment.
    • Four papers on graph-based parsing (using LOS graphs) have been put on MyCourses under "Readings." (Condon MSc + Hu PhD theses + 2 conference papers).
  • In-Class Project Discussion Assignment:
      Before class on Wednesday, submit a deck of 5 slides (in .pptx, Keynote or .pdf) to the "Group Discussion Materials" dropbox. The presentation should contain the following:
      1. Title slide (identifies students in the group, approach taken)
      2. Algorithms for Classification, Segmentation, and Parsing
      3. Features for Classification, Segmentation and Parsing
      4. Example(s) of training formulas recognized correctly.
      5. Example of an interesting/confusing/informative cases where recognition is incorrect (Note: these can be specific examples, or confHist output).
      We will discuss these as a group in-class on Wednesday and possibly a bit on Friday. Groups will sit together and present their slides from their seats, in the context of a class discussion about progress on the 3rd project.
  • Our last class will be next Monday.
  • Please fill out the course evaluation, and let me know what I can keep and change to make the course better for students in the future.
  • We will continue our discussion of parsing this week.

Week 13

  • !! Extension: Assignment 4 is now due on Saturday. This is to make sure that people presenting papers on Friday have a bit more time to finish the Assignment.
  • Project 3 is due next Friday.
  • The last round of paper presentations will be given in-class on Friday.
  • Please fill out the on-line course evaluation. This course is the way it is largely because of feedback from students. I'd be grateful if you could take the time to fill out the course evaluation, and let me know what worked and what you would change for future students in the course.
  • We will talk about HMMs and parsing techniques this week.

Week 12

  • Assignment 4 is due next Friday.
  • Project 3 is due in two weeks (Friday of Week 14).
  • Extension (the 2nd): Project 2 is now due Thursday at 11:59pm.
  • The final round of research paper presentations will be given next Friday.
  • We will finish up our discussion of segmentation, discuss the second project, and start our discussion of parsing this week.

Week 11

  • Extension: Project 2 is now due on Monday April 9th at 11:59pm.
  • The first round of paper presentations on Parsing will be given in-class on Friday.

Week 10

  • The second round of research paper presentations will be given this Friday. Papers for the first round of parsing papers next Friday will be posted this Friday.
  • Assignment 3 (project prep) is due Friday at 11:59pm.
  • Project 2 is due next Friday.
  • We will continue our discussion of segmentation this week.

Week 9

  • Hope you had a good spring break!
  • Papers for the second round of research papers on segmentation have been posted. Groups should sign in to MyCourses to look at papers and sign up for Groups.
  • Project 2 and Assignment 3 will be posted. A3 is due next Friday, and Project 2 in two week's time.
  • We will finish up our discussion of CNNs and start talking about segmentation this week. Read Casey and Lecolinet's nice survey on character segmentation! This is up on MyCourses.

Week 8

  • The first round of segmentation papers will be presented on Friday.
  • Read the paper on segmenting characters in OCR systems by Casey and Lecolinet (available through MyCourses).
  • Proj. 1 is due Monday night at 11:59pm.

Week 7

  • Deadline Extension: Project 1 is now due Monday night. Submit project materials using your Project Group dropbox.
  • The first set of segmentation papers will be posted this Friday. The groups that presented in the second group for the Classification papers will go first.
  • This week we'll try to finish up AdaBoost, have a look at Random Forests, and start taking a look at Convolutional Neural Networks.
  • Read Sections 2.3.5 through 3.3.5 of the Criminisi paper on Random Forests. The paper also includes introductory material, and material about applications of Random Forests for the XBox Kinect. Also read (at least) pages 1-17 of the original Convolutional Neural Network paper by LeCun et al., and have a look at the LeNet5 demo page (paper and link are available through MyCourses).

Week 6

  • The second round of Classification paper presentations will be given in-class on Wednesday.
  • Project 1 is due next Friday (Wk 7).
  • Read the two .pdfs from the Boosting text, posted on MyCourses (Chs. 1 and 3.4).

Week 5

  • Assignment 2 is due this Friday at 11:59pm. Please follow the submission instructions for the assignment carefully.
  • The second round of research papers will be in-class next Wednesday. For those presenting, make sure to create a Group in MyCourses, and to select your paper (first-come, first-serve) in the Doodle poll:
  • Project 1 has been posted, and is due Friday of Week 7.

Week 4

  • This Friday we will have our first round of paper presentations related to Classification in-class. Presentations will be given by groups of three students (with one group of two). Details are provided in the write-up on MyCourses.
    • Groups need to choose (any) group number and add their names through MyCourses. Click on the "Groups" link for the course, and add your name to the appropriate "Classification Paper Presentations" group. The first member will need to choose a PClass-[number] arbitrarily to create the group.
    • Link to choose paper for your group is here. List all last names for your group members when you make your seletion. First come, first serve, and your group may not change their selection, so talk to your group members first.
    • Four groups will present this Friday, and five groups will present next week.
  • This week we will finish up Bayesian Decision Theory, and talk about dimensionality reduction (PCA + LDA).

Week 3

  • Assignment 1 is due Friday, Feb. 2nd at 11:59pm.
  • Reading: read 'DuinEtAlCh2,' which provides an introduction to Bayesian Classification.
  • Office Hours Friday: due to a conflict, I will be holding office hours on Friday from 11am-12:15pm.
  • The first round of paper presentations will be given in-class next Friday. Papers will be posted this Friday, and students will select them on a first-come, first-serve basis.
  • This week we will discuss Bayesian Decision Theory.

Week 2

  • Assignment 1 is due next Friday (Feb. 2nd) at 11:59pm.
  • We will move into talking about nearest neighbor and linear classification this week. An associated reading (from Hastie et al.) has been posted on MyCourses (see Ch. 1 and Ch. 2.1-2.3).

Week 1

  • Note: the course syllabus and schedule are now available (links above).
  • Slides providing an overview of Pattern Recognition have been posted on MyCourses.