2012 Projects
Application of Self Organizing Feature Maps(SOFM) to Classify Mathematical Curves.
Advisor: Dr. Zohdy
In this project, SOFM are to be trained to categorize presented mathematical objects in the form of families of curves. Optimal features shall be extracted from x-domain,frequency-domain as well as joint x-frequency spaces obtained by transformations from the curves.Testing of learned curve classes shall be done following the training and convergence. New contributions are to be attempted, such as new features from joint x-frequency space, applying multiple similarity norms,use of pattern correlations, history keeping functions,and more effective initializations and input sequencing.
Lecture Notes |
Secret Key Extraction based on Network-Layer Characteristics in Mobile Ad Hoc Networks.
Advisor: Dr. Shu
Classical key establishment mechanisms, such as the famous Diffie-Hellman key exchange scheme, rely on the computational hardness assumption in solving certain hard problems (e.g., the discrete logarithm problem). The validity of this assumption has been seriously challenged by the tremendous computation capacity of today's parallel computing technology. Meanwhile, the most recently proposed physical-layer key extraction schemes are subject to the limitation of single-hop point-to-point channels. In this project, we will investigate novel mechanisms of generating secret keys by utilizing network layer characteristics such as routing, traffic signature, and nodes mobility. Such a mechanism enables two or more communicating parties that are multiple hops apart to establish secret keys without the needs of meeting each other before communication or relying on the computational hardness assumption. This research covers the current state of the arts in networking, security, and privacy. The research outcomes may contribute to the next generation security paradigm for mobile ad hoc networks.
Lecture Notes |
Attacks in Wireless Cellular Network.
Advisor: Dr. Fu
With so many people inputting their private information into
smartphones, the privacy of the users is a growing concern.This
project presents various growing concerns about smartphone privacy
issues. Our research covers current concepts of networking, security
and privacy in 3G and 3GPP with respect to hardware, software, and
application implementations on contemporary smartphones. After
analysis of these recommended protocols, we will address ways to
compromise users' privacy rights, and potentially develop fixes to
these attacks.
Lecture Notes |
Advisor: Dr. Li
Prostate cancer is the most commonly diagnosed cancer in men and is the second leading cause of cancer-related deaths in men. Functional imaging techniques such as diffusion weighted MRI (DW-MRI) are emerging that makes use of differences in microvascular and cellular content of normal and tumor-containing tissues. Although functional MRI has been used in staging, use of these techniques in tumor prognosis and therapy response assessment for prostate cancer has not been exploited. The goal of this study is to generate automatic statistical map based on anatomical and textural features to detect tumor likelihood and classify tumor aggressiveness in a DWI.
Lecture Notes |
Contrasting Data Mining Classification Methods.
Advisor: Dr. Qu
In this project, we will investigate methods for medical error detection and clinical behavior analysis. The team will work to design a new logging system on the top of the existing acute pain system (APS) that the information including patient medical records, nurses daily rounding, inpatient/outpatient follow ups and surgery procedure will be captured and also the uses of the APS system. Upon the collected data, instance based learning methods will be deployed. Once we have these two functions implemented, we will be able to answer the following questions in collaborating with the doctors: 1) what are the main reasons to cause medical errors? 2) how to avoid medical errors? 3) who is the nearest neighbor(s) for a specified patient?
Lecture Notes |
