The Fall Colloquium Series will occur on selected Friday’s from 2-3 pm in Old Union 213E throughout the semester. The colloquium series is an informal presentation in which presenters can get feedback on their topic of interest, as well as establish potential research collaborations. The colloquium is open to faculty, staff, and students. The topics and schedule our outlined below. We look forward to seeing you there!
If you are interested in presenting your research interests this fall or next spring, please contact Bryan Hosack at bhosack@ilstu.edu
Oct. 2 Chung-Chih Li “Union Theorems in Type-2 Computation” in Stevenson 104
The union theorem indicates that, informally, almost all natural complexity classes at type-1 such as PTIME, PSAPCE, EXP-TIME, EXPSPACE, and so on, fit the precise definition of complexity classes given by Hartmanis and Stearns. In other words, according to the theorem, the rigorous definition of complexity classes in terms of computable resource bounds is indeed broad enough to encompass most natural complexity classes. However, when we lift the computation to type-2 using oracle Turing machines, the union theorem doesn't hold without further strengthening some necessary conditions. In our previous work, we prove a non-union theorem under a less considered cost model known as unit-cost model. In this paper, we emphasize on a more popular cost model known as answer-length-cost model and give a full treatment of this powerful theorem at type-2. We prove and disprove several nontrivial variations of the union theorem based on our framework.
http://www.itk.ilstu.edu/faculty/chungli/mypapers/T2Unions.pdf
Oct. 9 James Wolf “An Investigation into the Economic Consequences of Adding an Online Sales Channel”
We used a unique dataset of 30,533 off-lease vehicle auction sales (10,898 offline and 19,635 online) to examine the economic impact of adding an Internet sale channel to a firm’s existing channel portfolio. Our findings suggest that an Internet channel adds economic benefit in several ways. First, the data suggest that prices are significantly higher and that the dispersion of prices for vehicles sold is significantly narrower in the online channel than in the offline channel.
In addition, the price difference observed between vehicles sold via online and offline channels is greater than one would expect to see from reduced transaction costs alone. Our data suggest that there are significant differences in the prices obtained for vehicles across locations and that and that these differences are associated with the flow of vehicles between states with vehicles flowing from low cost states to higher cost states. Finally, we show that the range of online prices is narrower that the range of offline prices. These findings are consistent with the suppositions that online prices converge more closely toward a national price than offline prices and that the online channel may allow the firm to capture a portion of the arbitrage rents.
Oct. 23 Amit Shesh “Images as graphs for searching and processing”
Content-based image retrieval (searching for images without resorting to text tags, but based on their content) has been a hot topic beaten to death, and yet there is not satisfactory commercial product/search engine out there that searches for images of realistic resolutions (what digital cameras are capable of producing) using various visual search cues. This research is yet another attempt towards searching images based on their content based on visual search cues like sketching parts of images roughly, using a similar image, etc.
Recently expressing images as graphs has become very popular, because it allows us to pose many of the image problems as graph theory problems. I will discuss what I consider as the "latest" approach towards it, its limitations and how I hope a different approach may lead to a better/faster search. Along the way, the image representation may also facilitate solving other problems like cartoon rendering, automatic detection of object boundaries, etc., thus creating the possibility of a generic image representation that is able to handle different image processing and searching tasks.
Oct. 30 Mary Elaine Califf "But What Does It Mean? Using Machine Learning to Disambiguate Words"
One of the challenges involved in applying computational techniques to natural language documents is the inherent ambiguity of human languages. One source of that ambiguity is vocabulary: one word may have many meanings. Determining the meaning of given use of a word from context, or word sense disambiguation, could theoretically improve success in many language processing tasks, but turns out to be a challenging task of itself. This research is attempting to improve on the current state of the art in word sense disambiguation by combining a pattern matching rule learning approach that works well for finding local context rules with current techniques that use information about the sentence or paragraph as a whole.
Nov. 6 Yongning Tang “Evidential Reasoning based Network Fault Localization and Anomaly Diagnosis”
Uncertainty is a fundamental and unavoidable challenge in network fault and security management. Various probability models have been proposed and adopted to tackle this challenge. However, in many situations, probability models have their limit and are commonly not applicable. In this presentation, I will first introduce the concepts of evidential reasoning of Dempster-Shafer theory (DST) that can be viewed as an extension to probability theory. Then, I will show examples on how to apply DST to diagnose network anomalies based on my recent research and some related work.
Nov. 13 Glen Sagers “58 Percent Secure: Why Do So Many Wireless Networks Not Use Encryption?”
Despite the fact that 802.11 wireless security has been available for at least a decade, 42% of wireless implementations remain unsecured. What factors lead to this condition? Is socioeconomic status to blame? Are only businesses concerned with security? Has security improved over the last few years? This research maps wireless access point locations via GPS and investigates these questions.
Nov. 20 Abhijeet Jain “Windows 7: What’s new?”
Windows Vista is said to be the biggest mistake of Microsoft as far as Windows is concerned. But will Windows 7 turn the tide? Is it worth upgrading to Windows 7 or is it just a minor upgrade to Vista? Topics to be covered-
© 2009 Illinois State University • An equal opportunity/affirmative action university encouraging diversity.• Privacy Statement
