Celebrate the end of the semester! - June 2nd 2006

We will celebrate the end of the spring semester on Friday, June 2nd. Cake (from Angel Maid Bakery!) and beverages will be provided. Please RSVP by adding a comment to this post.

Date: June 2, 2006
Time: 4:30pm - 5:10pm
Location: 11th floor conference room

Tutorial: Bayesian Techniques for NLP

Title: Beyond EM: Bayesian Techniques for Human Language Technology Researchers
Date: 24 May 2006
Time: 9am - noon
Location: 4th floor conference room

Expectation Maximization (EM) has proved to be a great and useful technique for unsupervised learning problems in speech and language processing. Unfortunately, its range of applications is limited either by intractable E- or M-steps, or by its reliance on the maximum likelihood estimator. The natural language processing community typically resorts to ad-hoc approximation methods to get (some reduced form of) EM to apply to NLP tasks. However, many of the problems that plague EM can be solved with Bayesian methods, which are theoretically more well justified. In this tutorial, I discuss Bayesian methods as they can be used in natural language processing. The two primary foci of this tutorial are specifying prior distributions and performing the necessary computations to perform inference in Bayesian models. I focus on unsupervised techniques (for which EM is the obvious choice), but discuss supervised and discriminative techniques at the conclusion with pointers to relevant literature.

Depending on one's inference technique of choice, the math required to build Bayesian learning models can be difficult. Compounding this problem is the fact that current written tutorials on Bayesian techniques tend to focus on continuous-valued problems, a poor match for the high-dimension discrete world of text. This combination makes the cost of entrance to the Bayesian learning literature often too high. The goal of this tutorial is to provide sufficient motivation, intuition and vocabulary mapping so that one can easily understand recent papers in Bayesian learning that are published at conferences like NIPS, and increasingly at ACL. In addition to the standard tutorial materials (slides), this tutorial is accompanied by a technical report that spells out all the mathematic derivations in great detail, for those who wish to start research projects in this fields.

This tutorial should be accessible to anyone with a basic understanding of statistics. I use a query-focused summarization task as a motivating running example for the tutorial, which should be of interest to researchers in natural language processing and in information retrieval. Additionally, though the tutorial does not focus on speech problems, those attendees interested in graphical modeling techniques for automatic speech recognition might also find the tutorial of interest.

Event Summaries

Experiences in Commercializing Research - March 30 2007

AI Grads hosted a panel where Craig Knoblock, Kevin Knight and Ed Hovy spoke about their expriences commercializing their research, and they gave great advice based on their histories. All of the speakers agreed that it is more important that a business idea solves a useful problem rather than just being really cool technology. The speakers also spoke about raising money and protecting your intellectual property. Another constant theme was the importance of hiring good business people such as lawyers and executives. The speakers explained that research companies can use SBIR and STTR grants to raise money, which was really insightful. We ate C & O Italian food and a tasty cake. 37 people (not including the speakers) and 1 dog attended this event, making it one of best attended events yet.

AI Grads lunch with alums - February 16th 2007

We had Greg Barish from Fetch Technologies, Radu Soricut from Language Weaver, and Marcelo Tallis from Teknowledge come in and discuss the transition to working in small business research setting. They pointed out some interesting differences between research in industry and academia, such as the necessity of weighing the potential of research ideas so as to keep a high level of productivity. They also described differences in writing proposals for SBIRs as compared to educational research grants. Marcelo also showed a really cool demo of integrating a deductive logic engine with a spreadsheet application like Microsoft Excel. We had about 27 people enjoying pizza and ice cream cake.

"AI in Games" a discussion by industry professionals - May 31st 2006

We had industry speakers from Savage Entertainment, Heavy Iron, and Pandemic Studios come and talk about their experiences with AI in games at their respective companies. Topics included how to simulate AI in games ("More Artificial then intelligent"), resource constraints, trade-off between expert AI and a fun user experience ("If your smart AI always kills me when I come into a room, that's not a game, its a torture chamber"), and working at game companies ("I make games for a living how bad can it be?"). We had 31 attendees including students and researchers. We served Johnnie's Pizza and soft drinks. (Pictures from the event)

LaTeX Document Formatting Tutorial - May 10th 2006

We had Matthew Michelson and Martin Michalowski present a tutorial on formatting documents using LaTeX . Matt gave a brief overiew of LaTeX and why we should use it. He followed with some "advanced" LaTeX tips and finished by showing some tools for Windows, demonstrating a couple of interesting features of each. Martin showed the main tools available for Mac, focusing on TeXShop in particular and presenting some of its cool features. The seminar also included an open discussion of specific questions related to LaTeX. About 20 people attended the event and empanadas, salad and cake were served.

Life After Graduate School - March 29th 2006

We had three speakers from our division at ISI present their views on different aspects of obtaining jobs after graduate school. The speakers were : (1) Prof. Kevin Knight, (2) Dr. Mark Moll, and (3) Dr. Yu-Han Chang. All three speakers had different backgrounds and perspectives. Prof. Knight talked about what he looks for when he hires graduates, while Dr. Moll talked about his experiences as a post-doc. Dr. Chang talked about how he got hired after graduating. About 30 people attended the event, which had Chipotle burritos for lunch.

Seminar Lunch: "AI in Games" a discussion by industry professionals

Do you like games? Do you want to know how professional game developers use/view AI? Then come to the seminar lunch planned for May 31st! This one will be about AI in video games and our speakers will be people in the industry! Of course, we are also providing free lunch. So come and learn what it takes to make a game fun, eat some food and talk about AI. As we confirm speakers, we will update this post. As usual, please RSVP to this event by adding a comment to this post.

Also, unlike past events, this event is open to anyone at ISI, so please encourage your fellow students and researchers to attend the event.

Confirmed Speakers:

• Savage Entertainment makers of such games as Star Wars: Battlefront II (for PSP) and Golden Eye: Rogue Agent.
• Heavy Iron Studios , part of the THQ family, makers of such games as The Incredibles!
• Pandemic Studios , makers of such games as Full Spectrum Warrior and Mercenaries 2!

Wednesday May 31st at Noon in the 11th floor conference room

WWW practice talk

I'll be giving a practice talk of "Towards Content Trust of Web Resources" (by Yolanda Gil and me) this Wednesday (May 17th) at 12:45pm in the 11th floor CR. The talk will be 25 minutes. Your attention and/or feedback, if you have time, would be very much appreciated!

Latex information

I was thinking this could be a post for people to put some useful Latex snippits, style files, etc. I might as well get this started by posting the USC thesis style file. You can find it here. The style file is actually pretty good at explaining things, but the main commands are:

1. \usepackage[final,thesis]{USCthesis} %Look in the style file for the different options, there are lots of them for things like qual, final, etc.
   
2. \title[]{blah}
3. \author{}
4. \committee{Prof 1\\*
   Prof 2\\*
   Prof 3}
5. \majorfield{COMPUTER SCIENCE}
6. \submitdate{Month 2005}
7. You use \chapter{chapter name} instead of \section{}

Lastly, to use the bibliography you need to somethign like:
\begin{singlespace}
\references[Reference List]{plain}{refs}
\end{singlespace}

Note that refs is the name of the bibtex file. I hope this is useful.

LaTex tutorial tools

Hey all,

These are links to the different tools I am going to discuss during the LaTex tutorial. You can download them (they are all free), and play with them to get a sense of how they work:

Miktex -- This is latex for windows. If you are on a windows box, you need this to use latex.

TeXnicCenter -- This is a LaTex IDE. For the new people, this will get you up and running quickly. You can click on icons to insert different Latex commands such as figures or symbols. For the more experience users, this IDE really saves you time in managing your TeX. For example, you can build your documents and display them all in one-click.

LaTable -- This tool lets you build tables visually and then copy the underlying LaTex code to paste into your document. Complicated tables are really painful to write, and this tool makes it a snap!

TeXAide -- This is a visual equation builder. It works similarly to Microsoft's equation editor. You pick symbols and such and build up your world-changing formula, then you can copy the LaTex that generates the formula and paste it into your document. This is awesome since equations in LaTex can be nasty.

A list of tools I will be talking about is in the comments section of this post.
-- Martin

Postdoc positions available

My dad is a professor at the University of Ottawa and he is looking for postdoc candidates for a medical triage project. The short description states:

The MET Research Group has received new funding for health informatics research entitled Triage Support at the Point of Care: Methodologies for Ubiquitous Systems in the Emergency Department. We are looking for dynamic graduate fellows (doctoral and postdoctoral) to join our multidisciplinary team for a period of 1-2 years starting in September 2006.

More information about the project can be found at its website and more specifics about the postdoc position can be found here. Let me know if you have any questions or simply contact my dad, he's a nice guy....

Latex Tutorial on May 10th 2006

The ai-grads taskforce is hosting a Latex tutorial on May 10th 2006. The tutorial will take place in the 11th floor conference room at 12 noon. Free lunch will be provided (empanadas). If you're interested, please RSVP by posting a reply to this post.

About Latex:
LaTeX is a high-quality typesetting system, with features designed for the production of technical and scientific documentation. LaTeX is the de facto standard for the communication and publication of scientific documents. Find out more at: http://www.latex-project.org/