Objectives of the Course and Preliminaries

Slides:



Advertisements
Similar presentations
Critical Reading Strategies: Overview of Research Process
Advertisements

Critical Reading Strategies: Overview of Research Process
Issues in developing narrative structures Postgraduate writing, seminar 7 John Morgan.
1 CSI 5387: Concept Learning Systems / Machine Learning Instructor: Nathalie Japkowicz Objectives of the Course and Preliminaries.
1 ETR 520 Introduction to Educational Research Dr. M C. Smith.
CS/CMPE 535 – Machine Learning Outline. CS Machine Learning (Wi ) - Asim LUMS2 Description A course on the fundamentals of machine.
Logistics: –My office hours: T, Th 4-5pm or by appointment –Class Web page:
1 CSI5388 Data Sets: Running Proper Comparative Studies with Large Data Repositories [Based on Salzberg, S.L., 1997 “On Comparing Classifiers: Pitfalls.
CSCI 347 – Data Mining Lecture 01 – Course Overview.
1 CSI 5388: Topics in Machine Learning Instructor: Nathalie Japkowicz Objectives of the Course and Preliminaries Course Webpage.
“Knowing Revisited” And that’s how we can move toward really knowing something: Richard Feynman on the Scientific Method.
Objectives of the Course And Preliminaries 1. 2  Instructor: Dr. Nathalie Japkowicz  Office: SITE  Phone Number: x 6693 (don’t rely.
40551 Logic Synthesis Optimization Dr. Yaser M. Agami Khalifa Fall 2004 Lecture # 1.
Experimental Evaluation of Learning Algorithms Part 1.
ScWk 242 Course Overview and Review of ScWk 240 Concepts ScWk 242 Session 1 Slides.
Network Simulation and Testing Polly Huang EE NTU
HE 520: Higher Education Laws and Regulations Unit One Seminar Pre-Seminar Welcome to HE 520: Higher Education Laws and Regulations, Unit One Seminar Seminar.
Most of contents are provided by the website Introduction TJTSD66: Advanced Topics in Social Media Dr.
1 CSI 5387: Concept Learning Systems / Machine Learning Instructor: Nathalie Japkowicz Objectives of the Course and Preliminaries.
Data Mining Practical Machine Learning Tools and Techniques By I. H. Witten, E. Frank and M. A. Hall Chapter 5: Credibility: Evaluating What’s Been Learned.
Research Word has a broad spectrum of meanings –“Research this topic on ….” –“Years of research has produced a new ….”
FNA/Spring CENG 562 – Machine Learning. FNA/Spring Contact information Instructor: Dr. Ferda N. Alpaslan
Course Overview Stephen M. Thebaut, Ph.D. University of Florida Software Engineering.
Statistics Critical Thinking in Intro Stats Roger Woodard.
CS & CS ST: Probabilistic Data Management Fall 2016 Xiang Lian Kent State University Kent, OH
Russell & Jamieson chapter Evaluation Steps 15. Evaluation Steps Step 1: Preparing an Evaluation Proposal Step 2: Designing the Study Step 3: Selecting.
Introduction to CSCI 1311 Dr. Mark C. Lewis
Dr.V.Jaiganesh Professor
Data Science Credibility: Evaluating What’s Been Learned
Requirements for the Course
Writing a Critical Summary of an Article or Paper
March 13, 2014 RS and GISc Institute of Space Technology
Welcome to Computers in Civil Engineering 53:081 Spring 2003
COMP 135: Human-Computer Interface Design
Introduction to CS Senior Design Project I / II
ACIS 3504 Accounting Systems and Controls
Parts of a Lab Write-up.
Seminar will begin at top of the hour
CSI 3120 Concepts of Programming Languages
The Basics of Literature Reviews
Geology I: Earth’s Interior(ERTH-1100)
Data Mining: Concepts and Techniques Course Outline
Sequencing Writing Assignments
Introduction to CS Senior Design Project I / II
Introduction to CS Senior Design Project I / II
Sequencing Writing Assignments
This Class This is a graduate level spatial modeling class in natural resources This will be one of the most challenging classes you’ll probably take You’ll.
Introduction to Statistics for Business Application
Evaluation and Its Methods
Tips for Reading a Scientific Research Article and Preparing your
COMS 161 Introduction to Computing
CHAPTER 9: Producing Data— Experiments
PHYS 202 Intro Physics II Catalog description: A continuation of PHYS 201 covering the topics of electricity and magnetism, light, and modern physics.
Physics 100 Energy Spring 2011.
And a short comment on note taking
Introduction to Comparative Effectiveness Course (HAP 823)
MRKT 485 BUSINESS-TO-BUSINESS MARKETING
ACIS 3504 Accounting Systems and Controls
CAP 6412: Advanced Computer Vision
Seminar DKV Desi Dwi Kristanto, M.Ds. Week 6.
Psych 231: Research Methods in Psychology
Introduction to CS Senior Design Project I / II
Evaluation and Its Methods
Introduction To CPSC 233 James Tam Event driven software
Designing Your Performance Task Assessment
Evaluation and Its Methods
Introduction To CPSC 233 James Tam Event driven software
Mastery Assessment in Teaching Statistics
GSP 470/570 Advanced Geospatial Analysis and Modeling
BISC 830 Community Ecology and Macroecology Course Logistics
Presentation transcript:

Objectives of the Course and Preliminaries CSI 5388: Topics in Machine Learning: Performance Evaluation for Classification Objectives of the Course and Preliminaries Course Webpage (including Syllabus): http://www.site.uottawa.ca/~nat/Courses/csi5388_2008.html

Some Information Instructor: Dr. Nathalie Japkowicz Office: STE 5-029 Phone Number: 562-5800 x 6693 (don’t rely on it!) E-mail: nat@site.uottawa.ca (best way to contact me!) Office Hours: Monday, Wednesday 1:00pm-2:00pm or by appointment Extra Seminars: TAMALE Seminars, (invited talks on Machine Learning and Natural Language Processing)

Why a Course on Performance Evaluation for Machine Learning?

The Issue Designing any kind of algorithms without testing it properly is useless: How do we know that what the algorithm is doing has anything to do with what we expect it to do? Performance evaluation for Machine Learning is particularly difficult: Machine Learning is based on the principle of induction  Even if the algorithm does well on the data it was trained on, how do we know that it generalizes to data it hasn’t seen before? Since data is scarce, we don’t know what kind of data the algorithm will see in the future. How can we, then, test any Machine Learning Algorithm?

Current Typical Solution Select a large number of UCI domains Train and test your new algorithm and those used for comparison using 10-fold cross-validation, on each domain. Use accuracy to evaluate performance. Average the results of the 10 folds for each algorithm on each domain. Run paired t-tests to make sure that the differences observed between the various algorithms are significant. If they are, claim victory, and publish happily! If not, play around with your algorithm, and run your experiments again, until you get satisfaction. Oh, and also, average the results obtained for each algorithm on every domain, to get a global idea of how your algorithm performs.

Is this solution acceptable? I don’t know… It makes some sort of intuitive sense, but I never truly investigated it. I simply made sure to apply it rigorously after my professor told me it was important to do so! However, various parts of the procedure have been challenged by researchers and practitioners inside and outside of the field of Machine Learning.

What are Some of the issues that came up? Basing our decisions on the metrics we typically use (e.g., Accuracy) is problematic. The data sets we use are not necessarily representative of those our algorithms will be applied to. The statistics we rely on is misleading: We may break the assumptions made by the statistical tests we use; Our sampling strategies may be inappropriate. As a result of all these issues, our results may not mean what we believe they do.

Purpose of the Course The purpose of the course is three-fold: First, it consists of gathering enough background knowledge in statistics, on metrics, and on experimental evaluation, in general, to understand the nature of the criticisms. Second, it consists of presenting some of the solutions that have been suggested to remedy the problems. Third, it consists of training the students to think critically about evaluation issues so as to design their own solutions.

Useful Reading Material Machine Learning References Machine Learning, Tom Mitchell, McGraw Hill, 1997. Data Mining, Witten, I & Frank, E., Morgan-Kaufmann, 2006. Introduction to Machine Learning, Nils J. Nilsson (available (free) from the Web) Statistics Reference StatSoft, http://www.statsoft.com/textbook/stathome.html Research Papers See the papers listed on the Web site

Objectives of the Courses: To introduce advanced topics in Machine Learning evaluation. To introduce the students to the careful reading, presenting and critiquing of individual research papers. To introduce the students to background research in a subfield of Machine Learning evaluation: finding appropriate sources (some giving broad overviews, others describing the most important approaches in the subfield), organizing the knowledge logically, presenting the knowledge to the class. To initiate the students to formulate a research problem and carrying this research through.

Format of the Course: A mixture of lectures and student presentations on various topics in the area of classifier evaluation. The student presentations will review and criticize recent specialized research papers written on the more general topic previously introduced by the lecturer. The last week of the term will be devoted to student project presentations.

Course Requirements: 12% 18% 30% 40% Percent of the Final Grade 6 paper critiques in which the student will critically and comparatively discuss the content of 2 or 3 research papers on the weekly theme. 3 critical and comparative in-class presentation of a research paper 3 assignments Final Project: - Project Proposal - Project Report - Project Presentation 12% 18% 30% 40%

List of Topics Review of Machine Learning's main concepts Current approaches for the evaluation of Machine Learning and their shortcomings Evaluation Metrics I: ROC Analysis / Cost Curves. Evaluation Metrics II: Non-Traditional Metrics, Evaluation through Projection, Combination of metrics Functional Elements of Statistics for Machine Learning 1 & 2

List of Topics (Cont’d) Sampling, Bootstrapping, Randomized Methods Putting it all together: Error Estimation of Machine Learning Algorithms Data Sets: Pros and cons of data repositories, simulated data sets Model selection, Theory, Practical Recommendations

Project (See Project Description on Course Web site) Research Project including a literature review and the design and implementation of a novel evaluation schemes, or the comparison of various evaluation schemes. Projects Proposal (3-5 pages) are due the week after the break. Project Report are due on the last day of classes Project Presentations will take place on the last week of classes Start thinking about the project early!!!!!