RIT Department of Computer Science CSCI-737-01 Pattern Recognition (Fall 2013)

Week 16

• Assignment 5 is due on Wednesday.
• Project 3 is due Wednesday, Dec. 18th.
• Project 3 Presentation Schedule:
  • Monday: Zack and Chris; Zack and Erika; Justin; Ravdeep and Piyush
  • Wednesday: Kevin and Ankan; Maulik and Himanshu; Wei and Fan; Joe

Week 15

• Project 3 has been posted, and is due Wed. Dec. 18th.
• The last set of research paper presentations will be given in-class on Wednesday.
• Assignment 5 has been posted. It is due next Wednesday at midnight.
• An additional research paper by Chou, which describes the Inside-Outside algorithm used to learn probabilities for stochastic context-free grammars has been posted on MyCourses (under "Research Papers").
• Presentation Schedule, Week 16:
  • Monday: Zack and Chris; Zack and Erika; Justin; Ravdeep and Piyush
  • Wednesday: Kevin and Ankan; Maulik and Himanshu; Wei and Fan; Joe

Week 14

• Project 2 is due Monday evening (11:59pm).
• Due to the holiday, there is no class on Wedensday or Friday this week.
• The last set of paper presentations (on parsing math notation!) will be given in-class Wednesday of next week. The poll has been posted; please sign up ASAP if your group will be presenting.
• The course schedule has been updated to reflect the holidays.
• Please complete the on-line course evaluations.

Week 13

• UPDATE: Project 2 is due Monday of Week 14 at 11:59pm.
• Project 3 will be assigned next Monday.
• The last three research papers for presentation (on parsing math notation) will be posted by Friday evening. They will be presented Monday of Week 15.
• Reading: Readings from Duda, Hart and Stork on string matching and parsing have been posted on MyCourses.

Week 12

• Research papers on parsing for presentation this Friday have been posted. Remaining papers will be assigned to groups on Wednesday by the ionstructor.
• Update: Assignment 4 will be posted on Wednesday evening.
• Project 2 is due next Friday at 11:59pm.
• Assignment 3 has been graded; a marked-up copy of submitted .pdfs are available through the Dropbox in MyCourses.

Week 11

• Bonus points for project 1 will be announced later this week.
• The first set of research papers on parsing will be posted on Friday. Students are again expected to present papers in groups of two.
• A reading on Hidden Markov Models has been posted. Slides from lecture and the reading are both available through MyCourses.

Week 10

• The second set of segmentation paper presentations will be given in-class on Friday.
• Project 2 has been posted on MyCourses.
• The visualization of the F-measure and a Precision-Recall graph shown in class on Wednesday is available here.
• The video illustrating character segmentation and recognition used to break CAPTCHAs is available here. There is also a Scientific American story about this.

Week 9

• Project 1 is due Friday at 11:59pm.
• Please see the notes posted under the Grade for the Segmentation Paper Presentations. These notes apply to all students in the course.
• Slides on k-means and fuzzy k-means clustering have been posted on MyCourses under "Class Notes."
• Student disk space quotas for CS accounts have been increased to 1GB. Recall that both LgEval and CROHMELib have been installed on the CS systems, at the locations:
  • /usr/local/dcs/versions/CROHMELib-0.1.12
  • /usr/local/dcs/versions/lgeval-0.2.11
  An overview of these applications and the CROHME dataset are available online: LgEval, CROHMELib and CROHME Data Overview.
• Here is the CROHME InkML File Viewer.

Week 8

• Project 1 is due next Friday.
• Reading for Wed of Week 9: a survey of character segmentation methods has been posted under "References" on MyCourses. We will discuss this paper in class starting Wednesday next week.
• Readings: readings on clustering from Duda, Hart and Stork's "Pattern Classification" have been posted; see also Jain et al's clustering survey (located under "Research Papers" on MyCourses), which provides a readable introduction to clustering and clustering applications.
• Research Papers (Segmentation, Part I): we will have research paper presentations in-class on Friday. The papers have been posted on MyCourses.

Week 7

• Assignment 3 has been posted. Extension: it is now due Tuesday at 11:59pm.
• Research Papers for Presentation Next Friday: will be posted online by Friday evening.
• Reading: notes/slides and a reading from Bishop's textbook on LDA and PCA have been posted on MyCourses.

Week 6

• Project 1 has been posted on MyCourses. It is due Friday of Week 9.
• Reading: a reading for Wednesday on Random forests has been posted (available through MyCourses - please read section 2.3 and Ch. 3). For Friday, notes/slides and a reading from Bishop's "Pattern Recognition and Machine Learning" have been posted on MyCourses.
• Project note: so that we can work with individual .inkml files, please use a randomized search to obtain folds with minimum average KL-divergence between the class distributions in each fold, and the global (i.e. actual) class distribution from all .inkml files (this will be discussed in-class on Wednesday). You will also want to avoid (possibly not completely) having no instances of a class in each fold (i.e. avoid having classes with 0 samples in a fold, where possible).

Week 5

• Schedule note, regarding Columbus Day (Week 8): we will have class on Tuesday rather than Monday, due to the holiday.
• Project 1 has been posted. The project will be completed in teams of two, and is due before 11:59pm on Friday of Week 9.
• Reading for Wednesday: please read the posted section from Boosting Ch. 3.4, and pages 303-309 of Ch. 10 (on boosting multi-class classifiers).
• Our second set of research paper presentations on classification will be given in-class on Friday.
• Project 1 will be posted on Friday. Projects will be completed in teams of two students. Note that the research papers being presented on Friday are related to Project 1 (copies of all papers presented are available through MyCourses).

Week 4

• Assignment 2 is due on Friday at 11:59pm.
• Reading for Monday: Chapter 1 of Freund and Schapire's Boosting (through MyCourses).
• Research papers for presentation next Friday will be posted and a sign-up web link sent out by end-of-day on Friday.
• Reading: look at C4.5 Ch. 4 on pruning for Friday's class (available through MyCourses).
• The schedule has been updated - Project 1 will be assigned next week, and be due Monday of Week 9.
• Grades for Assignment 1, and for the research paper summaries last Friday are available through MyCourses. You should see a marked-up version of the .pdf file that you submitted for assignment 1 in the dropbox and/or grading MyCourses pages.
• Research papers for the second set of paper summaries next Friday will be posted online later this week.

Week 3

• Please read C4.5 chs. 1 and 2 before class on Monday of next week (available under Content/References on MyCourses).
• Assignment 2 has been posted. It is due Friday of Week 4 at 11:59pm.
• On Friday, we will have our first batch of research paper presentations. Students should look at the abstracts and quickly scan the papers before class on Friday, and come with questions about them.

Week 2

• Assignment 1 is due Friday at 11:59pm.
• Classification papers for presentation next Friday have been posted. A description of expectations and grading of the presentations is available through MyCourses, under Content/Assignments. The doodle poll for selecting papers is located here. Students who do not present a paper next Friday will need to present a paper in Week 5.

Week 1

• Assignment 1 has been posted. Related reference materials and data are available through MyCourses. The assignment is due Friday of Week 2 at 11:59pm.
• Reading (for Wednesday, Week 2): read the chapter "Detection and Classification," from Classification, Parameter Estimation and State Estimation (available under "References" on MyCourses). Update: the "nuts and bolts" data set has been provided on MyCourses, so that you can try writing your own code for some of the examples shown.
• Reading (for Wednesday): please review Hastie Ch. 1 and Ch. 2.1-2.3 (on Least Squares and Nearest Neighbor classifiers), and the introductory course notes (slides).