Current Announcements:
o Due to an unavoidable off-campus meeting, Dr. Lathrop’s office hours will end at 11:30am on Wednesday, 6 Nov. Please come to office hours on that date before 11:30am, or send email to schedule an appointment anytime.
o Prof. Lathrop’s office hours for Weds, 16 Oct, are canceled due to the ICS Faculty Panel on Improving Your Grad School Application, which you are obliged to attend. If you are conflicted, you are obliged to watch the video when it comes up on the SAO website.
o The seminar CS 200S is currently full, but we anticipate that it will be easier to get into in the winter or spring quarter (Fall quarter is just always a difficult/crazy quarter!). However, if you would like to enroll in INF 209S now, you may email the instructor and ask if she would provide you with an authorization code to enroll. There is space in the classroom now. You may also substitute one quarter of any other ICS research seminar (e.g.,
o On Wed., 16 Oct., meet NOON TO 1PM in DBH-2011, for the ICS Faculty Panel on Improving your Grad School Applications.
o Students with time conflicts: Watch the video on the SAO website. [link will appear later]
o The
time and room location for co-located ICS-90 faculty research lectures has been
changed to Wednesday 4:00-5:00pm, EH-1200 (building 308 on the UCI campus map).
o NOTE:
Some ICS-90 meetings will not involve faculty research lectures, and can be
skipped. See schedule as it
develops (available at the bottom of this URL). The
faculty research lectures are 4-4:30pm and 4:30-5pm, which you are obliged to
attend if possible. The time slot
5-5:20pm is reserved for advice to freshmen, which you may skip.
o Prof. Lathrop’s office hours on Wednesday, October 9, will end at noon due to a department faculty meeting that begins at noon.
o Current announcements will appear here, at top-level, for quick and easy inspection.
William Strunk, Jr., and E. B. White, The Elements of Style, MacMillan, New York.
Robert A. Day (ed.), How to Write and Publish a Scientific Paper, Oryx Press, 1998.
Joseph M. Williams, Style: Ten Lessons in Clarity and Grace, Harper-Collins, 1989.
Dale Carnegie, The Quick and Easy Way to Effective Speaking: Modern Techniques for Dynamic Communication, Pocket Books, New York, 1977.
Robert M. Woelfle (ed.), A New Guide for Better Technical Presentations, IEEE Press, 1992.
The following represents a preliminary syllabus. Some changes in the
lecture sequence may occur due to earthquakes, fires, floods, wars, natural
disasters, unnatural disasters, or the discretion of the instructor based on
class progress.
Background Reading and Lecture Slides will be changed or revised as the
class progresses at the discretion of the instructor.
Week 1:
Wed., 2 Oct., Introduction to the ICS Honors Program.
Homework
#1, due Wed., 10 Oct.:
Find and read a research paper by an ICS Faculty Member.
o
Make sure the paper
already will be indexed in the literature (published at least 6 months ago in a
major venue).
o
Bring your paper to class on Wed., 9 Oct. (Students with time conflicts: email its title and authors to Dr.
Lathrop)
o
Bring
your paper to MRC-164 on Wed., 23 Oct. (Time conflicts: bring
it to the alternate library orientation, TBA)
o
If you already have a
research advisor and topic, this research paper must be from an entirely
different research area.
o
The intent is to broaden your exposure
to ICS research.
Week 2:
Wed., 9 Oct., Introduction to (1) ICS
Research Areas, and (2) the Scientific and Technical Research Literature.
o
Bring your ICS faculty research paper to class on Wed., 9
Oct.
Homework
#2, due Wed., 17 Oct.:
Summary, Point, & Question about your paper. [Instructions; Guide]
o
Email your 1-page Summary
to Dr. Lathrop by Wed., 16 Oct.
Background material:
Literature overview. [PDF;
PPT]
Technical writing. [PDF]
Week 3:
Wed., 16 Oct., Meet
NOON TO 1PM in DBH-2011, ICS Faculty Panel on Improving your Grad School
Applications.
o
Students with time
conflicts: Watch the video on the SAO website later. [link will appear here]
View the US
Bureau of Labor Statistics chart of education, income, and unemployment (click
here).
Homework
#3, due Wed., 23 Oct.:
Identify
five grad schools you might apply
to. For each school, identify one
professor there with whom you might work.
o
Email your
list to Dr. Lathrop by Wed., 24 Oct.
o
The reason for this requirement is so that when you write your
Statement of Purpose for each school, you will be able to mention at least one
professor *from that school* who
interests you.
o
Some professors will do a global string search for their name
through all of the applicant's Statements of Purpose, and then preferentially
read those applications that mention their name.
o
Thus, if you want to improve the chances that your application
actually will be read seriously, it is helpful to you to mention names of
professors at that school who interest you. Of course, if you mention a long list of
names it will be obvious spam, so mention no more than one to three. Also,
briefly mention something personal that attracts your interest about each,
again to indicate that it isn’t just impersonal spam.
o
These remarks apply mainly to Ph.D. applications. M.S. applications usually do not receive
as much carefully individual scrutiny as do Ph.D. applications.
o
Even if you currently do not
intend to apply to graduate school, you still must attend the Panel and do the
homework. If you ever change your mind later in
your career, after you know more than you do now, you will want to be properly
prepared.
Week 4:
Wed.,
23 Oct., Meet in MRC-164
at the Ayala Science Library (to right as you face library from Aldrich Park).
Introduction to Searching the Research Literature.
o Bring your ICS
faculty research paper to MRC-164 on Wed., 17 Oct.
o
Your challenge: Find your paper in the
literature databases using the online search tools.
·
Pretend you didn’t know it existed beforehand, but need to
find it --- Treasure Hunt!
Students with time
conflicts: I will try to arrange an alternate library
orientation, TBA.
Homework #4, Due Wed., 30 Oct.:
Identify three interesting ICS
research areas, and three interesting professors in each.
[Instructions]
o
Email your
1-page Summary to Dr. Lathrop by Wed., 30 Oct. (Deadline extended since hyperlink was broken; please
turn it in quickly.)
o
The point to this exercise is to show you that you have very
broad interests, and that you can find many ICS professors who might sponsor
interesting research.
o
If you already have a
research advisor and topic, all three research areas must be entirely
different. The intent is to broaden your exposure
to ICS research in general.
Week 5:
Wed., 30 Oct., Meet in ICS-259 rest of quarter.
How to Give a Successful
Technical Talk.
Homework #5, Due Wed., 7 Nov.:
Identify
one interesting ICS professor you might work with.
[Instructions]
o
Email your 1-page
Summary to Dr. Lathrop by Wed., 7 Nov.
o
The point to this exercise is to show you how to learn more about
an ICS professor before you approach them for research.
o
If you already have a
research advisor and topic, this ICS professor must be from an entirely
different area. The intent is to broaden
your exposure to ICS research in general.
Background material:
Example
talk: Intelligent Systems and Molecular
Biology. [Short
version for ICS-90]
Example archived student talks.
Week 6:
Wed., 6 Nov., How to Find and Join a Research Group.
No
more homework! Work hard in your
other courses!
Week 7:
Wed., 13 Nov., Responsible Conduct of
Research.
On Being a Scientist Video. Authors: Committee on Science, Engineering, and Public Policy, National Academy of Sciences, National Academy of Engineering, and Institute of Medicine
Responsible Conduct of Research. Presentation Material Credit: On Being a Scientist: A Guide to Responsible Conduct in Research, Third Edition, The National Academies; Michael Kalichman, Director, UC San Diego Research Ethics Program; Said Shokair, Director, UCI UROP; ICS Honors Program by Rick Lathrop, Director, ICS Honors Program.
You learned the basics in
kindergarten: Don’t lie,
don’t cheat, don’t steal.
*
Don’t lie: Don’t make
up false data or otherwise falsify results.
*
Don’t cheat: Don’t
misrepresent your data as something it is not or as better than it is.
*
Don’t steal: Don’t use
the words or data of others without proper credit and citation; don’t
plagiarize.
More complicated cases may require you to
consult for advice and disclose conflicts of interest so they can be managed.
When in doubt, seek advice and disclose conflicts.
Background
Material:
“Introduction to the Responsible Conduct of Research,” Nicholas H. Steneck with David Zinn, Office of Research Integrity (ORI), U.S. Department of Health and Human Services.
“UC Conflict of Interest Policy,” UC Business and Finance Bulletin G-39, UC Executive Vice President – Business Operations.
“Integrity In Research,” UCI Office of Research Administration.
“UC Office of Ethics, Compliance & Audit Services.”
Week 8:
Wed., 20 Nov., Research Intellectual Property
(IP) and Entrepreneurship.
Background
Material:
UC
Patent Policy, and Patent Acknowledgment.
Week 9:
Wed.,
27 Nov., Student Practice Technical Talks.
10:00-10:25
--- Drake Tetreault
Title:
Influence Analysis Through
Twitter Connectivity Mapping [PDF]
Abstract:
Social
media sites such as Twitter increasingly dominate the information dissemination
market which was previously the realm of television and newspaper. Analyzing
trends in social media is made made more difficult
than analyzing traditional media because of the ability of social media
participants to subscribe to as many information sources as desired, increasing
the potential number of inter-user links from linear in the number of users to
quadratic. For my research project I will implement a web crawler that continuously
queries Twitter use data from that site's public API and organizes the data
into a two dimensional influence map that depicts the relationships and
importance of various Twitter accounts. This representation will facilitate
analysis of the predominant information sources available to the modern
internet user.
10:25-10:50
--- Gio Borje
Title:
Combinatorial
Game Representation and Analysis of Snort [PDF]
Abstract:
The game
of a Snort is a partisan, combinatorial game where solving which player has a
winning strategy is a PSPACE-complete problem. We analyze the game of Snort by
describing a state space prove that the game is
unfair. However, rather than solving the game for general graphs, we
demonstrate generalized winning strategies for a few common families of graphs
such as star graphs, path graphs and cycle graphs.
Week 10:
Wed., 4 Dec., Student Practice Technical Talks.
10:00-10:25
--- Jay Tolentino
Title:
Crowdsourcing
and Aggregating Visual Design Critique [PDF]
Abstract:
Crowdsourcing
has been growing rapidly both in its efficiency and potential in managing and
leveraging microtasks. In particular, there is the
possibility of exchanging detailed, in-depth feedback with crowdworkers
of varied demographical backgrounds. In the case of visual designers, feedback
in real life is often limited to a designer's circle of acquaintances. The CrowdCrit systems developed by the
Human-Computer Interaction Institute of Carnegie Mellon utilizes the
opinions available through crowdsourcing in order to provide more varied
feedback for the designers. The biggest challenge of the project is understanding how to scaffold the system in order to
yield helpful feedback for designers. After extensive prototyping of the
critique aggregation interface, the full system was finished the summer of 2013
and a multistage user study was conducted to better understand the system 's overall helpfulness for designers.
10:25-10:50
--- Jonathan Stroud and Igii Enverga
Title:
Ensemble Learning and the Heritage
Health Prize [PDF]
Abstract:
The
Heritage Health Prize is awarded to the team that develops the most accurate
model for predicting how long a person will spend in the hospital during the
next year, given a set of historical medical data. A successful prediction
algorithm could revolutionize the healthcare industry, as it could preemptively
identify patients with a high risk of hospital admission and reduce the number
of unnecessary hospitalizations. In this work, we implement and optimize a
number of commonly used machine learning algorithms for this problem and
compare their performances to that of an ensembling
algorithm that combines the predictions from all models. We also explore the
effects of feature design and selection methods on these individual models and
their consequent ensemble. We conclude that an ensemble is significantly more
effective than any individual model, reaching position 88 out of 1309 on the
final Heritage Health Prize leaderboard.
