CS464 Proposal Presentation Instructions

User Manual: Pdf

Open the PDF directly: View PDF PDF.
Page Count: 1

DownloadCS464 Proposal Presentation Instructions
Open PDF In BrowserView PDF
CS-464 Proposal Presentation Instructions
Fall 2017
Instructor: Mehmet Koyutürk (koyuturk@cs.bilkent.edu.tr)
TA: Iman Deznabi (iman.deznabi@bilkent.edu.tr)
The purpose of the project is to increase your knowledge about machine learning and get hands on
practical experience. Any project in the machine learning field that can be completed in the given time
frame can be proposed. You will work in teams of 5 students. The project can involve applying known
methods to solve an interesting question, or it can involve coming up with a new methodology to solve
an existing problem on an existing data set. For example:
- You can propose a problem that requires significant work for feature extraction. In this case,
you can use off-the-shelf implementations of machine learning algorithms to perform training
and classification since the core of the work you will propose will be feature extraction.
- You can propose comparing the performances of different machine learning algorithms on a
new problem. In this case, you can use off-the-shelf implementations of multiple (say three or
four) machine learning algorithms and comprehensively evaluate the performance of these
algorithms on your data, by systematically taking into account all possible factors (different
values of the parameters of learning algorithms, data size and dimensions etc.) and multiple
performance criteria.
- You can propose improving the performance of a given machine learning algorithm by making
an extension that you think will fit well in the context of a specific application. In this case, your
work will focus on algorithm development/implementation and evaluation (you need to
systematically compare the performance of your implementation to what is already available,
using a fair assessment scheme).
Proposal Presentation (on Tuesday, 10/10 and Friday 10/13)
During your proposal presentation, you are expected to provide concrete answers to the following
questions:
-

What is the problem you will solve?
What is the approach you will take? Here, “approach” refers to methods for feature extraction
(if needed), model building (including feature selection if necessary), learning algorithm. You
can propose multiple methods and compare them.
- What data is available for you to use in this project? What are the main characteristics (number
of samples, number of features etc.) of the data? If you will collect data, what are your goals?
- How will you evaluate your approach? How will you use your data to evaluate your approach?
- What is the amount of work you will devote to each component of the project. It is expected
that each student will devote a total of 10 hours to the project (excluding the preparation of
presentations and reports). How will the work be divided among team members?
During the presentation, you will have 10 minutes, including questions and answers. Thus you may
want to limit your presentation to 6 slides (1 slide for problem description, 2 slides for approach, 1
slide for data description, 1 slide for evaluation, and 1 slide for work statement).

Proposal Document (due Tuesday, 10/17)
The proposal document will also contain answers to these questions, and it will be limited to one page.
The document should also address any questions and concerns that were raised during your
presentation. You may want to do this explicitly (by stating the concern that was expresssed and
providing a response) or implicitly (by providing a description that addresses the concern).
Source Exif Data: 
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.5
Linearized                      : Yes
Author                          : Mehmet
Comments                        : 
Company                         : 
Create Date                     : 2017:10:03 13:58:36+03:00
Modify Date                     : 2017:10:03 13:58:36+03:00
Source Modified                 : D:20171003105826
Subject                         : 
Tagged PDF                      : Yes
XMP Toolkit                     : Adobe XMP Core 5.6-c015 84.159810, 2016/09/10-02:41:30
Metadata Date                   : 2017:10:03 13:58:36+03:00
Creator Tool                    : Acrobat PDFMaker 17 for Word
Document ID                     : uuid:d72e20c4-0953-4816-be65-8b0c2c4f0838
Instance ID                     : uuid:34f9b87f-5fd5-4086-9aa8-b1e0698811df
Format                          : application/pdf
Title                           : 
Description                     : 
Creator                         : Mehmet
Producer                        : Adobe PDF Library 15.0
Keywords                        : 
Page Layout                     : OneColumn
Page Count                      : 1
EXIF Metadata provided by EXIF.tools

Navigation menu