Differences

This shows you the differences between two versions of the page.

--- projects [2018/08/29 15:37] – lesperan
+++ projects [2018/09/18 15:44] (current) – lesperan
@@ Line 1: / Line 1: @@
 ~~NOTOC~~
 ====== Proposed Projects   ======
+====== Heart and Voice Signal Processing For Sleep Apnea Detection ======
+**Supervisor:** Hossein Kassiri
+Sleep apnea is a neurological condition where the patient’s breathing is interrupted during sleep, which can lead to excessive daytime sleepiness, irritability, and morning headaches. Left untreated, it increases the risk of numerous serious health issues, including high blood pressure, irregular heartbeat, and stroke. An estimated 860,000 Canadians reported being told by a health professional that they have sleep apnea, but many more people likely remain undiagnosed.
+This project will be done in collaboration with a Toronto-based medical device company that is working on the development of a miniature wearable device for sleep monitoring and apnea detection. The project will be conducted in two phases. In the first phase, the goal is to study and evaluate the efficacy of using various sensory signal (i.e., biomarker) in sleep apnea detection. A short list of candidate biomarkers are selected as a result of literature review, and their importance in apnea detection is evaluated using pre-recorded labelled online datasets. In the second phase, a signal processing algorithm for sleep apnea detection is developed that uses the selected biomarkers from Phase I as its inputs. In addition to detection, the algorithm is intended to evaluate the episode’s severity, frequency, and patient-specific pattern.
+The collaborating company will develop the hardware that contains the biosensors identified by the students and will be responsible for firmware development to control and extract data from the sensors. The company will also develop a mobile app that displays the data obtained, and produces test units that can be used by real human testers.
+**Required skills:** General CSE408x prerequisites, good programming skills, good math skills, knowledge of C or MATLAB programming languages.
+====== Patient-specific epilepsy seizure detection using machine learning ======
+**Supervisor:** Hossein Kassiri
+Accurate detection of neurological disorders such as epilepsy seizures using brain signals requires algorithms that can adjust themselves from patient to patient and from time to time. Data-driven algorithms using machine learning provide us with such patient-specific tools that allow for processing the data recorded from the brain and decide whether or not it can be classified as an abnormal activity for a specific patient.
+This project is aimed at development and optimization of such a machine learning algorithm for early detection of epilepsy seizures. The student will use pre-recorded labeled data from an online standard dataset to train, cross-validate, and test the algorithm. The algorithm is later optimized for computational resource consumption.
+**Required skills:** General CSE408x prerequisites, good programming and math skills, knowledge of C or MATLAB programming languages, basic knowledge of machine learning and pattern recognition (EECS 4404 or similar).
+====== Mobile app development for interfacing with a brain implant ======
+**Supervisor:** Hossein Kassiri
+Our group of researchers in ICSL are working towards implementation of an implantable closed-loop microelectronic system-on-chip (SoC) for recording and processing of EEG signals to be able to detect and treat various neurological disorder patients.
+Online monitoring and control of brain signals is one of the key features that can help the clinician significantly. Currently an electrical engineer has to be on-site to setup the system and adjust it upon clinician’s request. Therefore, a simple GUI that can be operated by a non-engineer enables the clinician to control the process entirely. The project’s goal is to develop a mobile application that is capable of receiving signals and sending commands through a Bluetooth link. Also the app should be capable of displaying the recorded signal.
+The student is expected to have experience with mobile app development (iOS or Android).
+**Required skills:** General CSE408x prerequisites, mobile app development (iOS or Android) experience.
+====== Biometric access control ======
+**Supervisor:** Amir H. Chinaei
+Biometrics as a means of access control has been previously studied and
+found to be a popular choice for guaranteeing authentication and
+authorization. This includes: iris, voice, face, fingerprint, and hand
+geometry recognition. Biometric features possess an if-and-only-if
+relationship. `
+This project investigates a solution that enables emergency medical
+technicians to have simple and fast, remote and token-free, as well as
+privacy-preserved and reliable access to patients’ medical information. The
+idea is to provide the technicians with a mobile system including a
+biometric reader device, through which they gain access to necessary
+attributes of patients’ electronic health record using the patient’s
+biometrics, such as fingerprint. Reliability is employed by exploiting the
+uniqueness of a person’s fingerprint as a means of access control as well as
+by precision of biometric reader devices, such as fingerprint scanners.
+Privacy of patients is preserved by enforcing an arbitrary privacy policy.
+In such system, the only required tokens are some biometric information;
+patients need not to carry with them an additional token—such as a health
+card, driving license, etc., to receive the service.
+**Required skills:** General CSE408x prerequisites, passion for reading and
+writing critiques, experience with embedded devices and programming.
+====== Decentralized access control using ABAC and application plugins ======
+**Supervisor:** Amir H. Chinaei
+This project aims at designing and developing software prototype that helps
+to advance security of information and privacy of users in web
+applications--with the means of decentralized administration. The project
+focuses on data security and user privacy in popular application domains
+such as health information systems and social media. The goal is to devise
+and integrate comprehensive mechanisms towards decentralized
+privacy-preserved administration models, with web-based and user-friendly
+interfaces and implemented by application plug-ins. The methodology is to
+first analyze the privacy risks that exist in current web applications such
+as in social media applications. A role-based only approach to data privacy
+is not sufficient because role-assignment must be managed centrally to be
+efficiently implemented. Thus, the assignment of a role to a particular
+object for a particular subject is administrated by the system so there is
+no way for a user to assign access rights beyond those defined by some
+central authority. If no matching role is defined a priori, the user is left
+with the difficult task of (1) attempting to create a new role (leading to
+role explosion); (2) being overly restrictive in selecting a “close” role
+(reducing system utility and limiting access below what is actually needed);
+or (3) selecting a role that is too general, leading to potential breaches
+in security or privacy violations. Instead, the proposed approach lies in an
+attribute-based solution in which users can create objects, modify the
+content, and control the access. It is obvious that a minimal set of
+application-dependent corporation policies, to which all users comply, is
+needed to avoid anarchy.
+**Required skills:** General CSE408x prerequisites, passion for reading and
+writing critiques, experience with web developer tools and plugin
+development.
+====== Privacy preserving constructs for programming languages ======
+**Supervisor:** Amir H. Chinaei
+Programming languages suffer from the lack of sufficient features that serve
+privacy preservation and ensure data security in cyber spaces. Consequently,
+many applications that have been developed by these languages are vulnerable
+with respect to data security and/or user privacy. For instance, XML, a
+popular underlying language for many web based applications, needs
+enhancements onto its access control component, XACML, by adding features to
+support privacy preservation. Developing extensions for data presentation
+formats such as XML, XAML, JSON, etc. to make them more suitable for a
+secure and trustworthy cyber space is an ongoing problem. Obviously, any
+mechanism to define privacy policies in such documents needs to be supported
+by the query languages too. One approach is to study XACML and XQuery and
+investigate minimal modifications such that they support ad-hoc attributes
+of privacy preservation. To measure feasibility and effectiveness of the
+modifications, one may need to develop prototypes to the proposed solutions
+and conduct experiments.
+**Required skills:** General CSE408x prerequisites, passion for reading and
+writing critiques in the areas related to SQL engine and data exchange
+formats such as XML and JSON.
+====== Building healthcare question answering systems ======
+**Supervisor:** Amir H. Chinaei
+In this project, the prospective student(s) will get the opportunity to
+build question answering systems for healthcare or education applications.
+The project involves simpler and more complex tasks: data collection using
+open source search engines, data annotation, building machine learning
+models, evaluation of the built system, etc. The students will involve in
+one or several tasks depending their interests and skills.
+**Required skills:** General CSE408x prerequisites, programming (preferably in
+Python and Java) to interact with large datasets for experiments and
+analyses.
+====== Machine learning for real time access control ======
+**Supervisor:** Amir H. Chinaei
+As artificial intelligence methods continue to improve, so do the
+opportunities of their applications on different industries. Finance and
+healthcare databases include sensitive data that certain people should
+access in a certain time and location. For example, health information (such
+vulnerability to certain diseases) are information that only a healthcare
+practitioner should access in a given time and location. Consider when a
+practitioner needs to access an attribute of a patient during a surgery, or
+when an employee tries to access to a particular data remotely, or when
+employees change their positions over time in a company. In these and many
+situations, granting and revoking data access needs to be updated online and
+real time. To this end, there is a demand for intelligent systems for online
+and real time access control. In this project, the prospective students will
+get the opportunity to apply machine learning techniques on access control
+data to learn patterns of users accessing data and to automatically learn
+access control rules.
+**Required skills:** General CSE408x prerequisites, programming (preferably in
+Python and Java), passion for reading and writing critiques in the area of
+data security and machine learning.
 ====== Detecting and Visualizing anomalies in dynamic networks ======
-**Supervisor:** Arjun An
+**Supervisor:** Aijun An
 In this project, the goal is to detect and visualize anomalies in dynamic graphs. Graphs are powerful tools to model networks and relationships between their entities. Many of the real-world graphs are dynamic in which nodes and edges are being added and deleted over time. Some of the dynamic network examples include social networks and computer networks. Anomalies are any deviation from the normal. For example, in a computer network, communications are normal over time until a single machine is attacked by a large number of other machines at a specific time point. We work on finding time points that an anomaly occurs and nodes, edges or subgraphs that are responsible for this anomalous behavior. This project focuses on finding and visualizing anomalies in a stream of graphs. The student will first use a visualization tool to visualize the network over time. This helps in better understanding the changes in the structure of the network. There are various visualization tools available to use for this purpose. In the second phase, the student will implement a graph anomaly detection method and combine it with the visualzation program so that the graph stream can be monitored through visualization and anomalies can be detected and illustrated in an online fashion. This project is a part of a collaborative project with IBM and the student will work closedly with a PhD student in the project.
@@ Line 108: / Line 257: @@
-====== Design and development of a low‐power backend PCB for brain neural signal processing and wireless data transmission ======
-**Supervisor:** Hossein Kassiri
-The student will be responsible for the design and testing of a printed circuit board (PCB) that receives brain electrophysiological signals from multiple recording sites; analyze, decode, and organizes them; and transmit them through a wireless link to a handheld (e.g., cellphone) or stationary (e.g. a laptop) device. The board must be designed while the strict power and area budgets of a medical device taken into account.
-Once developed, the prototype along with the currently‐available wearable brain monitoring device will form a complete wireless wearable solution for point‐of‐care brain monitoring and diagnosis applications.
-**Required Background:** Solid knowledge of electrical and electronic circuits, proficiency in Verilog and MATLAB, previous PCB design experience is not a must but definitely is a plus.