Seminars

The information on this page is not completed and schedule is tentative. Please check back later for more information.

  • Title: Increasing Perceived Agency in Story-based Video Games
    Speaker: David Thue

    Date: Friday, February 1
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: Agency, being one's ability to perform an action and have some in?uence over the world, is fundamental to interactive entertainment. Although much of the games industry is concerned with providing more agency to its players, what seems to matter more is how much agency each player will actually perceive. In this presentation, we present a computational model of this phenomena, based on the notion that the amount of agency that one perceives depends on how much they desire the outcomes that result from their decisions. Using a structure for high-agency stories that we designed speci?cally for this intent, we present the results of a 141-participant user study that tests our model's ability to select subsequent events in an original interactive story. On average, players who experience event sequences as selected by our model report higher perceived agency than players who experience sequences that our model does not recommend.

  • Title: Security on Ultra-Low Power Devices
    Speaker: Qian Yu

    Date: Friday, March 15
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: Nowadays, there are many applications of the Ultra-Low Power Devices (ULPD). For many of these applications, some level of security is of utmost importance. However, applying security services on ULPD is not easy due to the hardware limitation and severely limited power resources. This presentation introduces a hash function and a secure data transmission scheme which can perform on ULPD. Firstly, an introduction to ULPD and related cryptography background will be covered. Secondly I will introduce an important stream cipher called RC4 and the reversible property of RC4 states. Thirdly, an RC4 based hash function and its applications will be described. Lastly, I will present an RC4 based secure data transmission scheme for ULPD.

  • Title: Constraint Satisfaction Problem Solving on a GPU
    Speaker: Ahmed Mobaraki

    Date: Monday, March 18
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: There are many problems in Computer science, especially in Artificial Intelligence, that are classified as constrained problems. A constrained problem consists of three main components: variables, values and constraints. Finding a solution for this kind of problem is very expensive in terms of time. Constraint Satisfaction Problems (CSP) provide a framework for representing these problems. CSP has been extensively studied for the past few decades, and researchers have proposed a variety of techniques for implementing constrained problem solvers based on the characteristics of the problem and its constraints. In this presentation I will talk about the implementation of a CSP solver on GPU (Graphics Processing Unit).

  • Title: Fault Tolerance in Multiagent Systems
    Speaker: Jordan Ubbens

    Date: Friday, March 22
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: Probabilistic multiagent systems use probabilistic inference to solve the problem of distributed interpretation. In a probabilistic multiagent system, each agent has knowledge of only a subset of the network domain and expresses its belief as a probability distribution over its subdomain. Agents communicate with each other by message passing to form a joint probability distribution representing the expected state of the domain. In this talk, we explore probabilistic multiagent systems and their underlying graphical representations. The issue of fault tolerance is introduced and a framework called risk propagation is proposed to handle unreliable or noisy agents.

  • Title: Introduction to Recommender Systems
    Speaker: Sojung Kim

    Date: Monday, March 25
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: Recommender Systems (RSs) aim to make a prediction about a user's preference. The objective of the systems is to reduce the information overload problem, and suggest items that the user had not yet considered but might find interesting. In this introductory presentation, basic RS concepts and techniques will be discussed. Working examples of finding similar users, ranking potential items, and recommending movies will also be presented. A brief introduction to the presenter's thesis direction, recommender systems as a research support, will be presented.

  • Title: Survey of Online Auction Systems
    Speaker: Tahani Alotaibi

    Date: Wednesday, March 27
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: In e-commerce, online auctions are important trading mechanisms that allow buyers to find their items of interest with less cost and in less time. Existing online auction systems come with different features (forward vs. reverse, single-round vs. multi-round, single-attribute vs. multi-attribute, open vs. sealed bids) as well as with different bidding protocols (English, Vickery, first-sealed bids, Dutch). In this talk, I will discuss in detail all of these auction features to assist buyers and sellers in making their decisions.

  • Title: Decision-theoretic shadowed sets
    Speaker: Xiaofei Deng

    Date: Wednesday, April 3
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: A shadowed set, proposed by Pedrycz, is a three-valued approximation of a fuzzy set. Given a pair of thresholds (a,b), an element whose membership grade equals to or is greater than a is elevated to 1, an element whose membership grade equals to or is less than ?is reduced to 0, and an element whose membership grade is between b and a is put into a shadowed region. A fundamental issue is the determination of thresholds for producing an optimal approximation. In the light of three-way decisions, this seminar introduces a decision-theoretic model of shadowed sets, in which the required thresholds are computed by minimizing decision cost.

  • Title: Intelligent Tutoring Systems Measuring Student Effort During Assessment
    Speaker: Peter Lach

    Date: Friday, April 5
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: Each tutoring system must face the problem of dealing with uncertainty involved in student interactions. Not all correct answers to the same question are the same. Some answers are produced by careful execution of thoughts using memorised facts, where the others are just pure guesses. Therefore there is a need to develop a tutoring system which takes into consideration the quality of an answer to produce a more detailed student model. In this presentation, I discuss an exploratory study of eye tracking technology which is used to modify the student model in Intelligent Tutoring Systems, during assessment, to create a more accurate estimate of a student's knowledge state. Using additional data about a student allows us to create a more precise model of the student knowledge state, which can lead to better adaptation of the learning instructions.

  • Title: An Experimental Approach to the Construction of Binary Decision Classes from Card Sort Data
    Speaker: Emad Almestadi

    Date: Monday, April 8
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: A card-sorting study of facial photographs was conducted with 25 participants who were asked to sort mixed raced photos into piles according to how they perceived similarity. If different people have different strategies for making similarity judgments, the differences should be visible in their sortings. A goal is to find one or more pairs of decision classes that best split the participants into groups based on these strategies. This talk discusses a computational approach that generalizes earlier work to find good decision classes within the sorting data. Experiments are described that led to a few good candidates which were then tested for accuracy of classifier using the Rough Set Exploration System.

  • Title: Compression and Analysis of Motion Capture Data
    Speaker: Kael Dow

    Date: Wednesday, April 10
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: The use of motion capture in video games is becoming more common and the mechanics of human motion lends itself to unique methods of compression. We will take a look at the current methods of compressing motion capture data and design considerations when creating compression and decompression techniques. Topics that will be covered will include key framing and principal component analysis, as well as techniques for compressing large databases of motion capture data by applying clustered principal component analysis and motion pattern indexing.

  • Title: Motion Detection of a Vehicle in a Video Stream
    Speaker: Alex Wang

    Date: Friday, April 12
    Time: 3:30pm - 4:20pm
    Room: CL 408

    Abstract: In this presentation, I will give an introduction to my research that deals with motion detection of a vehicle in a video stream. The input video stream is taken from a regular used vehicle wholesale auction house. In the video, there is not only the motion of vehicles, but also of people (including bidders, the auctioneer, and bus boys). Detecting moving vehicles will be very useful for dealing with auction arbitration problems. Currently, these problems are resolved by looking separately at the video record and the auction transcript archive. My work will help to index the video stream and map the video stream to the transcript. Based on a Frame Difference Analysis of the input video, a collection of convex hulls are generated for local dynamic regions. My work will identify and apply suitable pattern recognition techniques to reliably extract the subset of convex hulls that correspond to the vehicle in the scene. This work will also enable more intelligence in the business.



To Top of Page