B.Sc. The University of British Columbia, BC, Canada, 2005
M.Sc. Simon Fraser University, BC, Canada, 2009
Monday, February 20th, 2012 3:30 p.m. TASC1 9204 West
MONITORING TASK EXPERTISE WITH EYE TRACKING
Performance of a manual task involves coordination of the hands and eyes to execute a motor plan and achieve a desired goal. When the task includes manipulation of a tool or object, a certain level of expertise is needed to complete the task effectively. Traditionally, task expertise has been measured only by parameters of the task performance itself. More recently, researchers have begun to study the eye component of eye-hand coordination and agree that there are marked differences in gaze patterns exhibited by novice and expert task performers.
Many occupational tasks require special motor manipulation skills, thus prior or provisional training is needed which can be expensive and time-consuming. Like task expertise, training has normally emphasized development of motor control through feedback regarding actions performed while the role of the eyes has been ignored. Some studies have shown that the rate of motor skill acquisition for certain tasks can be accelerated by providing feedback on the trainee's gaze behaviors, suggesting the notion that having expert gaze patterns can aid the development of expert task performance.
This report surveys the techniques and analyses used in eye tracking studies for both direct-vision and image-guided manual tasks.
Ph.D. Depth Examining Committee
Dr. Stella Atkins, Sr. Supervisor
Dr. Bin Zheng, Supervisor
Dr. Ted Kirkpatrick, Supervisor
Dr. Lyn Bartram , Examiner
Wednesday, February 22nd, 2012 12:3 0 p.m. TASC1 9408
INVESTIGATING SOCIAL MEDIA CONTENT:
PROPOGATION, MONETIZATION, AND ENHANCEMENT
In the past decade, numerous significant social media sites have emerged in the Internet, especially video sharing sites (VSS), represented by YouTube, and online social networking sites (SNS), represented by Facebook and Twitter. Social media sites are known to have such major characteristics as rich user experience, dynamic content, and more important, the social network. Social media have drastically changed the content distribution landscape, and with no doubt, they have become an important part of people's everyday life.
The recent research works have largely studied these popular social media sites in the literature. In this proposal, we will further investigate the social media content distribution from two extended perspectives. On one hand, we will take an important step towards understanding the characteristics of video sharing propagation in SNS, based on large-scale data traces from RenRen Network. On the other hand, we will examine the long-term large-scale insight data of YouTube partner accounts, trying to discover more characteristics as well as understand the factors affecting the viewership.
Based on the results of the two measurement studies, we try to leverage VSS and SNS. In one case study, we will investigate users' personal preference of online video sharing from a questionnaire survey, and propose an appropriate overlay structure to accommodate the new application scenario. In another case study, we will identify the challenge of current social media content storage, and address this by introducing load-balanced migration to content cloud.
Ph.D. Thesis Proposal Oral Examining Committee
Dr. Jiangchuan Liu, Senior Supervisor
Dr. Jian Pei, Supervisor
Time:
11:00am - 12:00pm
Description:
Ph.D. THESIS PROPOSAL ORAL EXAMINATION
Mohsen Jamali
B.Sc: Sharif University of Technology, Tehran,Iran, 2005
M.Sc: Sharif University of Technology, Tehran,Iran, 2007
Monday, February 27th, 2012 11:00 a.m. TASC1 9204 West
RECOMMENDER IN SOCIAL NETWORKS
Recommender systems are becoming tools of choice to select the
online information relevant to a given user. Collaborative filtering
is the most popular approach to building recommender systems and has
been successfully employed in many applications. With the advent of
online social networks, the social network based approach to
recommendation has emerged. This approach assumes a social network
among users and makes recommendations for a user based on the
ratings of the users that have direct or indirect social relations
with the given user. As one of their major benefits, social network
based approaches have been shown to reduce the problems with cold
start users. In this research we propose novel methods to address
the recommendation problem in online social networks. The proposed
models consists of memory and models based approaches that exploit
the past ratings of users together with the social network among
user to predict the unknown ratings of users.
Ph.D. Thesis Proposal Oral Examining Committee:
Dr. Martin Ester, Senior Supervisor
Dr. Jian Pei, Supervisor
3D VIDEO STREAMING AND BROADCASTING: CHALLENGES AND APPROACHES
Three-dimensional (3D) media has recently gained a strong interest both in the research community and in the market. Various companies are introducing new 3D capable devices every year in consumer electronics shows. Many TV networks are starting to realize the potential of the 3D TV market and are launching dedicated 3D channels. The advances in 3D video acquisition and display technologies have paved the way for many emerging 3D applications, such as free-viewpoint video, 3D TV, and immersive teleconferencing. Such applications expand the user experience beyond what is offered by traditional media. In this report we survey recent developments in the fields of 3D video coding, transmission, display, and quality evaluation. We review some of the research works in the literature investigating adaptive and reliable transmission of 3D videos over wired and wireless networks. We also highlight the main challenges and future research directions.
Ph.D. Depth Examining Committee
Dr. Mohamed Hefeeda, Sr. Supervisor
Dr. Joseph Peters, Supervisor
Dr. Jiangchuan Liu, Examiner
Wednesday, February 29th, 2012 2:30 p.m. TASC1 9204 West
KERNELIZATION OF NP-HARD PROBLEMS ON SPARSE GRAPHS
Abstract: Under the assumption that NP is not equal to P, there are numerous natural problems for which super-polynomial running time is inevitable when complexity is measured in terms of the input size. It is while some inputs for such problems consist of sections that are easy to deal with and those that are more difficult. This sometimes make the problem computable in a time that is polynomial in the input size and exponential or worse in a parameter k. Hence, if k is fixed at a small value and the growth of the function over k is relatively small then such problems can still be considered "tractable" despite their traditional classification as "intractable". Parameterized complexity hence can be regarded as two-dimensional generalization of P vs. NP where while estimating the running time in addition to the overall input size n, the effects of a secondary measurement that captures additional problem-relevant information, is also taken into account. An effective approach in fixed-parameter algorithmics is that before starting a cost-intensive exact algorithm to solve a fixed-parameter tractable problem characterized by a parameter k, a polynomial-time pre-processing phase is executed to shrink the input data of size n to a smaller instance. The solution for the original input then can be reconstructed in polynomial time in n using a solution for the shrunk instance. The shrunk instance is called problem kernel. It is then hoped that the size of the problem kernel is upper-bounded by a polynomial in k, independent of n. The process is then called kernelization.
In this presentation, I review some of the techniques and frameworks introduced in the paramterized algorithms literature for kernelization of restricted classes of graphs. The main focus is on the classes of sparse graphs which admit small problem kernels.
Ph.D. Depth Examining Committee
Dr. Qianping Gu, Sr. Supervisor
Dr.Pavol Hell, Supervisor
Dr. Jiangchuan Liu, Examiner