COSC 7362—Advanced Machine Learning Fall 2015 ( Dr. Eick )

last updated: December 11, 3p

Basic Course Information

class meets: Mo/WE 1-2:30p in 204 AH
Instructor: Dr. Christoph F. Eick
office hours (573 PGH): MO 3:15-4:45p WE 2:30-3p
cancelled classes: We., October 21, Mo., November 16.
makeup class (for cancelled class): Fr., November 20, 9:45a-noon
class room: 204 AH

Course Summary

Syllabus COSC 7362 Fall 2015

Focus: One focus of the course is to learn how to read, summarize, present, review, and evaluate scientific papers. The papers that will be discussed during the course originate from the following areas: Anomaly Dection, Density Estimation, Deep Learning, Ensemble Learning, and Spatial-Temporal Clustering. Moreover, you will get some exposure to current developments and research in machine learning and related fields.

There will be also some more general discussion on machine learning research methodology and on how to conduct a scientific research project. Finally, there will learn and practice how to write abstracts, introductions, conclusions, paper reviews, and executive summaries.

Reasons to take the course: The course is a good preparation for master thesis and PhD dissertation research in the areas machine learning, data mining, databases, data analysis, and artificial intelligence. Moreover, seeing how well-known scientists present their research results will hopefully help you to do a better job in conducting your own research and in presenting your research results in your future publications.

Course Elements: 2 quizes, 4-5 paper walkthroughs, student presentations summarizing the content of machine learning papers, survey style lectures on some of the subfields covered in Fall 2015 and on (machine learning) reasarch methodology, 2 homeworks, discussion of papers and homework solutions, machine learning software demos.

Tentative Weights of the Different Components of the Course: Quizes: 46%, Presentations: 36%, Homeworks: 14%, Attendance and Class Particiaption: 4%.

COSC 7362 News

  • The final grades and all the scores are available now! Just click on the COSC 7362 Scores and Grades link below!
  • I enjoyed teaching the course and I already liky to wish you a happy, and successful year 2015.
  • Here is some feedback for the 3 homeworks: Abstracts, Conclusions, Introductions, H2 Reviews, Homework3.
  • 7362 Scores and Grades (as off December 11, 2015) (Solution Sketches Quiz1; Some Solution Sketches Quiz2.

    Papers Covered

    Overview Chapter Outlier/Anomaly Detection (for Paper Walkthrough on September 2, 2015; read before that date!!)
    Outlier Detection with Mixture Model (Student Presentation Paper; to be presented by Nguyen Pham on Sept. 16, 2015)
    Orair's et al.'s Paper on Distance-based Outlier Detection (Student Presentation Paper; to be presented by Kajol Agarwal on We., Sept. 23, 2015)
    Silverman's Classical Density Estimation Paper (for Paper Walkthough on September 9; read before that date!!)
    Kim and Scott's Paper on Robust Non-Parametric Density Estimation (companion paper to the paper, listed next!)
    Vandermeulen et al.'s paper on Robust Non-Parametric Density Estimation that was presensented at NIPS 2014 (to be presented on Sept. 28 by Khalil, Nacer)
    Survey Paper on Deep Learning Approaches (to be used for paper walkthrough on Sept. 30/Oct. 7)
    Target Propagation (an alternative approach to back propagation; to be presented by Lifeng Yan on October 5)
    Paper on Deep Convolutional Neural Networks (main source for Arthur's deep learning presentation on October 12, 2015)
    Papers on Trajectory Clusterings: Visual Analytics Approaches, Patition and Group Approach, Survey paper.
    Paper on Trajectory Classification using Region-based and Trajectory-based Clustering (main source for Kajol's presentation on Nov. 9/11, 2015)
    Paper on Trajectory Classification with HMM (main source for Ha's presentation on Nov. 11, 2015)
    Ensemble Learning Walkthough Paper
    Original ADABOOST Paper (main source for Yongli's and Nacer's presentation on Nov. 18, 2015)
    Dietterich's Classical Paper on Ensemble Learning (main source for Lifeng's presentation on Nov. 20, 2015)
    Paper on Stacked Generalization Techniques for kNN (main source for Tianxiao's presentation on Nov. 20, 2015)
    Paper on The Dynamics of AdaBoost (main source for Nguyen's presentation on Nov. 30)
    Paper on an approach for the Netflix Contest; used in Arthur's presentation on Dec. 2.
    Book Chapters on Sequential Prediction Problems and Recurrent Neural Networks, to be used in part in Kunal presentation on Nov. 30, 2015.

    Other Papers Worth Reading

    Scott's Outlier Detection Paper
    Paper Xiong et al. on Group Anomaly Detection which additionally introduces hierarchical Gaussian Models
    Density Estimation Techniques in R
    Clustering via Non-Parametric Density Estimation Techniques in R
    Rice University Article on How to Read Scientific Papers

    Class Transparencies

    First Lecture: Course Information COSC 7362
    Introduction Anomaly Detection (covered on August 26)
    Introduction to Parametric Density Estimation (covered on August 26/31)
    Introduction to Non-parametric Density Estimation (covered on August 31)
    Andrew Moore's Tutorial on (Gaussian) Mixture Models (will discuss a few slides of this tutorial; likely, on August 31)
    Research is Exploring and Overcoming Boundaries (covered on September 14)
    How to Read and Write Scientific Papers (update on October 14, 2015)
    Reviewing of Scientific Papers (covered on October 14/19)
    How to write the ‘Empirical Evaluation’ Section of Your Paper (covered in the lecture on October 19)

    Student Presentations and Other Useful Material for Review Purposes

    Nguyen Pham's Presentation on Anomaly Detection with GMMs
    Kajol's Presentation on Distance-Based Outlier Detection
    Ha and Nikoloas Presentation on Density Estimation Teachiques in R
    Khalil Nacer's Presentation Slides
    Khalil Nacer's Supplementary Material
    Kunal's Presentation on Back Propagation
    Kunal's backpropagation example
    Dr. Eick's Backpropagation Slides
    Lifeng's Presentation on an Alternative Backpropagation Approach
    Arthurs's Presentation on Conv. NNs
    Yongli's Presentation on Trajectory Clustering
    Tianxiao's Presentation on Trajectory Clustering
    Nikolaos's Presentation on Temporal Activity Clustering
    Kajol's Presentation on Trajectory Classification Relying on Region and Trajectory Clustering
    Ha's Presentation on Trajectory Classification with HMMS
    Yongli's and Nacer's Presentation of AdaBoost
    Lifeng's presentation covering Dietterich's article on ensembles
    Tianxiao Presentation dicussing a stacked generalization approach for kNN-classifiers
    Kunal's presentation on recurrent neural networks and sequential prediction problemsgeneralization
    Nguyen presentation on the convergence of AdaBoost
    Arthur's presentation on the Netflix Contest


    Translation number to letter grades:
    A:100-90 A-:90-86 B+:86-82 B:82-77 B-:77-74 C+:74-70
    C: 70-66 C-:66-62 D+:62-58 D:58-54 D-:54-50 F: 50-0

    Only machine written solutions to homeworks and assignments are accepted (the only exception to this point are figures and complex formulas) in the assignments. Be aware of the fact that our only source of information is what you have turned in. If we are not capable to understand your solution, you will receive a low score. Moreover, students should not throw away returned assignments or tests.