|
||||||
|
||||||
| Michigan State University CyberSecurity Initiative Workshop | |||
| Saturday, November 20, 2004 | |||
| Kellogg Center at Michigan State University | |||
| Big Ten Room C | |||
Synergistic Research Opportunities in Health Information Technology and Cybersecurity |
The complex issues associated with acquiring, processing, managing, transmitting and protecting electronic health information pose challenges to improving the health, safety and security of individuals, families and communities in the State of Michigan as well as in the entire country. A key aspect in this domain is the development of interoperable health information systems, an objective that requires the interaction of business, technical, clinical, administrative, legal, sociological, educational and public health experts. Michigan State University , joining with industry and health care community organizations as partners is uniquely positioned to undertake this objective. The theme of the Workshop is to explore the issues and opportunities related to this objective and to learn about some of the related issues such as safeguarding confidential information in a wireless world and regional and national health information exchanges. |
Workshop Program 8:30 a.m. – 9:00a.m. Registration, Continental Breakfast 9:00 a.m. – 11:00 a.m. Opening Remarks and Invited Presentations 11:15a.m. – 12:15 p.m. Research Poster Display and Informal Discussion of Current Research in Cybersecurity at Michigan State University 12:30 p.m. – 1:30 p.m. Working Lunch The Michigan State University Initiative: Synergistic Research Opportunities in Health Information Technology and Cybersecurity Opening Remarks and Invited Presentations Opening Remarks : Session Moderator : |
Kristy LaLonde |
| Policy Analyst |
| Executive Office of the President, Office of Management and Budget |
Abstract This presentation will offer an overview of some national initiatives in the areas of health information technology and cybersecurity. We will discuss how initiatives in each of these areas relate to one another, and how they fit in to the framework of current law and policy. In addition, we will discuss possible opportunities in an academic setting and at the state and local government level within the context of the E-government section of the President's Management Agenda and President's Executive Order on Health Information Technology. Ms. LaLonde currently serves as a Policy Analyst in the Information Policy and Technology Division at the Office of Management and Budget, within the Executive Office of the President. She works closely with Federal Agencies, including NIST, the Office of the National Health Information Technology Coordinator at the Department of Health and Human Services, and the Department of Homeland Security, to help coordinate and implement some of the ongoing national initiatives in the area of information technology (IT) management. Specifically, Ms. LaLonde's portfolio includes policy analysis and IT capital planning in the areas of health IT and cybersecurity. Prior to her work at the Office of Management and Budget, Ms. LaLonde was a Presidential Management Fellow in the Justice Management Division at the U.S. Department of Justice Headquarters. She holds a B.A. in Criminal Justice and a B.S. in Zoology from Michigan State University , and a Master's of Public Heath in Health Management and Policy from the University of Michigan . |
Manfred Tatzmann |
| Co-Chair, Michigan Department of Community Health Electronic Health Record Committee |
Abstract MDCH is faced with the challenge of managing five state hospital with an e-environment created in the mid 70's and early 80's. Patient care, medical charts, physicians orders, evaluations, and patient billing, are mostly paper driven and where electronic, not always uniform. Facilities have to interact with multiple spheres of activity and have record keeping requirements not found in the private sector. EMR systems developed for acute setting do not meet the unique needs of the behavioral health care environment. The presentation will describe these and other challenges faced and how DCH is working in a broader context of state government to improve the service delivery, the quality of patient care, standardize clinical protocols, reduce errors, while focused on cost reduction. Mr. Tatzmann has worked for the State of Michigan for 33 years. He accepted the current assignment as an avocation, while serving as the DCH State Traumatic Brain Injury Project Director. His interest in this field stems from his involvement in an early 70's project to develop a statewide computer information and referral network. Although trained in the business administration field, his career has included working for former Governor Milliken as his Human Services Director, key executive positions with the former Department of Mental Health, and developing a number of human service programs for the state. His work has been recognized by state and national organizations. Recently, DCH released the findings of a five-year study that was prepared under his direction on incidence and prevalence of traumatic brain injury in Michigan and the impact it has on the state Medicaid budget. This landmark study has been cited as a national model. |
David Ellis |
| Corporate Director of Planning and Future Studies, Detroit Medical Center Executive Director of the Michigan Electronic Medical Record Initiative |
Abstract The Michigan Electronic Medical Record Initiative (MEMRI) is a not-for-profit corporation formed to develop an extensible, standards-based, patient-centered electronic medical record (EMR) system, by linking disparate computerized patient record systems and clinical data repositories across independent providers throughout the state. The architecture for MEMRI is based on open, standards-based, modular, scalable and secure solutions from Sun and AccessPt Inc. It brings together, in near real time, information from various healthcare information technology sources and systems in a virtual place—a Web portal—for physicians, clinicians, patients, and others to access from a standard web browser. All information is encrypted and information being viewed is secured by access rules, rights, and privileges set by the participating facilities and enforced by identity and access management. This presentation will provide a high-level overview of the MEMRI technology architecture, focusing on the security aspects. Mr. Ellis is corporate director of planning and future studies at the Detroit Medical Center . A former "China watcher," he morphed into managing editor of a demographic journal, founder of a successful US regional Internet service provider, author ("Technology and The Future of Health Care," Jossey-Bass, 2000), publisher of the monthly online publication, "Health Futures Digest" (http://hfd.dmc.org/), and co-founder and executive director of the Michigan Electronic Medical Record Initiative. His role at the DMC is to help its physicians, nurses, and management be aware of the accelerating trends in health-related technologies and to position DMC to take advantage of the opportunities afforded by those trends to advance its goals and fulfill its mission to improve the health of the people it serves. He is a member of the World Technology Network, and was educated in England , Hong Kong and the United States , with degrees in business studies, Chinese and the information and communication sciences. |
Shaun Grannis, M.D. |
| Medical Informatics Researcher, Regenstrief Institute, Inc. and the Indiana University School of Medicine |
Abstract Because the information needed to answer important health research, management, and policy questions is usually scattered across many independent databases, secure methods accurately linking patient records from independent sources are critical to developing regional and national health information infrastructures. While valuable information can be gained from linked data sources, steps must be taken to insure patient confidentiality. This presentation will describe secure methods that use de-identified patient data. These methods were developed at the Regenstrief Institute using the Indiana Network for Patient Care (INPC), one of the nation's largest regional health information exchanges with nearly a half-billion coded results collected on three million patients over three decades. The genesis and evolution of this standards-based infrastructure will be presented, including success factors and lessons learned. Dr. Shaun Grannis is a Medical Informatics Researcher at the Regenstrief Institute, Inc. and Indiana University School of Medicine, where his research interests include exploring and overcoming the challenges of integrating data from distributed systems for use in health care delivery and research. Dr. Grannis developed a patient record linkage algorithm using cryptographically deidentified demographic data for use in distributed clinical data networks. The goal of the linkage algorithm is to maintain patient confidentiality while providing researchers with access to clinically meaningful data.. Dr. Grannis is also actively involved in bio-terrorism detection and syndromic surveillance. He helped craft guidelines for biosurveillance implementation and participated in research examining the value of over-the-counter drug sales for use in disease outbreak detection. He is currently involved in multi-year studies that explore several challenges facing disease detection and syndromic surveillance, including geographical de-identification, understanding temporal-spatial disease trends, and establishing syndromic surveillance data standards. He is the lead investigator for a 4-year project integrating data flows from all hospitals in the state of Indiana for use in clinical care, disease surveillance and clinical research. Dr. Grannis also maintains a clinical practice and is faculty in the Department of Family Medicine at Indiana University . After receiving his engineering degree from the Massachusetts Institute of Technology, Dr. Grannis spent two years in South America as an oil field analyst. He received his medical degree from Michigan State University 's College of Human Medicine in 1997, and subsequently completed a residency in Family Medicine. Prior to his research position at Indiana University and the Regenstrief Institute, Dr. Grannis was an instructor in the department of Business Information Systems at Central Michigan University . |
Research Poster Display and Informal Discussion of Current Research in Cybersecurity at Michigan State University Session Moderators : Dan Kim Robyn Mace
• Secure ID Allocation in Sensor Networks Hongbo Zhoju, Computer Science and Engineering; Dr Mutka, and Dr Ni (HKUST) Advisor: Dr. Matt Mutka, Computer Science and Engineering Globally unique ID allocation is not applicable in a sensor network due to the large number of nodes in the network, the limited bandwidth and the size of the payload. However, locally unique IDs are still necessary for nodes to implement unicast communications to save power consumption. Several solutions have been proposed for locally unique ID assignment in sensor networks, but they bring much communication overhead. This is not desirable due to the limited power supply in a sensor node. Combined with a directed diffusion communication paradigm, a reactive ID assignment scheme with security mechanisms is proposed. It defers ID conflict resolution until data communications are initiated and thus saves communication overhead. It is also resilient to packet loss and invulnerable to specific attacks from malicious nodes.
• Expose or Not? A Progressive Approach for Service Discovery in Pervasive Computing Environments Feng Zhu, Computer Science and Engineering Advisor: Dr. Matt Mutka, Computer Science and Engineering In pervasive computing environments, service discovery facilitates users to access network services by automating tedious manual configurations. When network services become pervasive, the number of service providers also increase dramatically. Because of the security and privacy concerns, network services are segmented by service providers. Existing service discovery protocols, however, do not address how to facilitate users to properly identify and authenticate with existing service providers. Without prudence, sensitive information may be exposed. Conversely, with prudence both users and service providers prefer the other party to expose sensitive information first. We identify that even among legitimate users and service providers, there are privacy concerns that may be expressed as a chicken-and-egg problem. We propose a progressive approach to solve the problem. Users and service providers expose minimal sensitive information in turn and identify necessary exposure during the process. Theoretical analysis, simulation and experiments show that our approach protects sensitive information with little overhead.
• Physical Layer Built-in Security Enhancement of Wireless Systems Using Secure Block Interleaving Qi Ling , Electrical and Computer Engineering Advisors: Dr. Tongtong Li, Electrical and Computer Engineering Dr. Jian Ren, Electrical and Computer Engineering The physical layer built-in information privacy of the conventional CDMA system, provided by pseudo-random scrambling, is far from adequate and can be improved by applying cryptographic techniques in the scrambling process. Motivated by the fact that after scrambling, chips spread from one symbol still cluster together and could be fragile to strong burst errors and fading effects, in this paper, a chip-level secure interleaving procedure is proposed to improve the system performance while enhancing the security measure. More specifically, the AES algorithm is combined with block interleaving. It should be noted that interleaving is in fact a special case of scrambling. Security analysis is provided to demonstrate the effectiveness of the proposed secure interleaving scheme under exhaustive search attack. Simulation examples are presented to illustrate the robustness of chip-level interleaving over channels with severe fading or strong burst errors.
• Joint Spatial-Spectral Image Watermarking Mahmood Al-Khassaweneh, Electrical and Computer Engineering Advisor: Dr. Selin Aviyente, Electrical and Computer Engineering With the advance of the Internet, the transmission, distribution and access to digital data have become easier than before. This advance brings with itself the challenge of protecting the content of digital data. This problem has generated the need for research in digital watermarking. Traditionally, the problem of digital image watermarking has been approached in the spatial and spectral domains . In this paper, a new approach that combines the two domains through joint time-frequency representation is considered. This new approach provides flexibility in determining the pixels to be watermarked and increases the data hiding capacity. An easy to implement watermark embedding algorithm and the corresponding detection algorithm are presented. The watermarking algorithm is tested under different attacks and is shown to be robust. • Watermarking Strategies for Authentication of Speech Transmissions Aparna Gurijala, Electrical and Computer Engineering Advisor: Dr. Jack Deller, Jr., Electrical and Computer Engineering A general formulation for speech watermarking through parametric modeling is presented. The watermark information is embedded in linear predictive coefficients of speech. The watermark embedding process can be interpreted as an FIR filter design problem. A particular example of watermark detector design is discussed and the false positive and misdetection rates are derived. Watermark robustness to channel noise and signal processing operations is analyzed and experimental results are presented.
• Teens and Online Privacy Elizabeth Taylor Quilliam, Advertising Advisor: Dr. Nora Rifon, Advertising As adolescents have increasingly embraced the Internet, threats to their online safety and privacy have multiplied. Risks include privacy scams, loss of personal information, identity theft and exposure to potential stranger interactions. For teens, the Internet can be an unsafe place. Like other Internet users, teens can adopt safe online behaviors and protect their privacy. This study applies concepts from Social Cognitive Theory and Protection Motivation Theory to adolescent online behavior, using a sequential, multi-method approach to investigate what teens know about protecting themselves online. The relationships between online self-efficacy, perceived risk of negative outcomes from online activity, perceived vulnerability to negative outcomes, and response efficacy are examined using both quantitative and qualitative techniques. The study uses interviews, focus groups and questionnaires to collect data from adolescents, parents and teachers. In the first phase, computer and business teachers at a Midwestern high school will be interviewed about their perceptions of their students' online behavior. Following the semi-structured interviews, focus group interviews will be held in a computer-equipped classroom at the high school. Students will have Internet access during the interviews, allowing the use of websites as stimuli. Utilizing findings from both sets of interviews, two self-administered questionnaires will be developed for completion by the focus group participants and by their parents. Incorporating online privacy protection into public school education is a tool available to policy makers. By better understanding adolescent motivations, behaviors and responses, appropriate message strategies can be developed and used to create high school curricula. One outcome of this study will be the development, with high school administrators, of such a curriculum. • Authorization Model in a Virtual Enterprise Yuan Zhang, Computer Science and Engineering Advisor: Dr. Moon-Jung Chung, Computer Science and Engineering We present a formal specification of the Layer-based Access Control (LBAC) model for virtual enterprises. In a virtual enterprise, workflow-related information is shared between participating units, which join or leave the virtual enterprise dynamically based on the running status of inter-organizational workflows. Access control inside each unit and between units is critical for a secure collaborative environment. To control redistribution of information among participating units is a major access control concern in virtual enterprises. In our model, an Access Layer is an “envelop” that hides access control details from the outside and provides formal interfaces to support access requirements exchanging between different Access Layers. Our approach introduces formal notations for modularized access control policy expressions, rules and an inheritance condition /c/ for guiding policy inheritance along (role, traditional/virtual organization, and process) hierarchies, routine for access request evaluation, and rules for policy conflict resolution.
• A Study of the Role of Web Assurance Seals in Online Consumer Trust Ying-Ju Lai, Telecommunication, Information Studies and Media Advisors: Dr. Dan Kim, Telecommunication, Information Studies and Media Dr. Charles Steinfield, Telecommunication, Information Studies and Media There is conflicting evidence as to the current level of awareness and impact of assurance seals services. This study examines consumers' awareness of assurance seal services and consumers perceived importance of assurance seals found on business-to-consumer (B2C) e-commerce websites. We further examine whether an intervention to increase consumers' knowledge of security and privacy threats, as well as informing them of the function of third party seal (TPS) providers increases perceived importance web assurance seals. We find that educating consumers about security and privacy threats, as well as the role of web assurance seals does increase awareness and perceived importance the seals, but consumers still are not likely to use these seals as an indicator of the trustworthiness of site. Moreover, there is no difference after the intervention in perceived security and privacy of a vendor's site, suggesting that consumers obtain these cues from other factors. After the intervention, the strongest predictor of perceived security remains consumers' assessment of the quality of the information provided on the site. However, the perceived privacy afforded by a vendor was no longer predicted by perceived information quality. • FISOD: A Robust Outlier Detection Algorithm for Multimodal High-Dimensional Data Sets Sohraab Soltanis, Computer Science and Engineering Advisor: Dr. Pang-Ning Tan, Computer Science and Engineering Outlier detection is the task of identifying observations whose characteristics deviate significantly from the rest of the data. Techniques developed for detecting outliers have been successfully applied to a variety of applications including network intrusion detection, fraud detection, and fault detection in mechanical structures. The key objectives of an outlier detection algorithm are two-fold: to ensure that the detection rate is high while maintaining a low false alarm rate. To achieve these objectives, we need to develop an accurate profile for the normal observations, which is a challenging task especially when multiple profiles are needed to represent the normal observations in data. In this project, we develop a new algorithm for detecting outliers called FISOD (Frequent Item Set based Outlier Detection), which uses an ensemble of closed frequent patterns to represent the normal profile. Unlike distance-based outlier detection schemes, this approach is highly suitable for finding outliers in sparse, high-dimensional data. To enhance the detection rate and reduce the false alarms, an ensemble-based strategy is employed. Our experimental results demonstrate that FISOD outperformed ORCA, a distance-based outlier detection algorithm, on several synthetic and real-world data sets. • Privacy Preserving Classification Jung-Eun Lee, Computer Science and Engineering Advisor: Dr. Pang-Ning Tan, Computer Science and Engineering Privacy issues are garnering wider attention particularly with the increasing interest in applying data mining tools to reveal interesting information about individuals. To balance the need for enabling the use of such tools while protecting against the disclosure of user sensitive information, privacy-preserving data mining techniques have been developed. Such techniques would apply a random perturbation method to sanitize the sensitive information first before applying existing data mining tools. However, one caveat of these techniques is that the models or patterns derived from the sanitized data tend to be less accurate than those derived from the original data. This poster presents an approach for privacy preserving classification using distribution shifting methods. Despite not knowing the original data, this approach attempts to recover information loss in the sanitized attributes by exploiting the relationships between the private and public attributes. Our preliminary experimental results suggest that classification accuracy can be improved using the proposed method while maintaining the privacy of users. • 3D Face Recognition Xiaoguang Lu, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering The performance of face recognition systems that use two-dimensional images depends on consistent conditions w.r.t. lighting, pose and facial appearance. We are developing a face recognition system that utilizes three-dimensional shape information to make the system more robust to arbitrary view, lighting and facial appearance. For each subject, a 3D face model is constructed by integrating several 2.5D face scans from different viewpoints. A 2.5D scan is composed of one range image along with a registered 2D color image. The recognition engine consists of two components, surface matching and appearance-based matching. The surface matching component is based on a modified Iterative Closest Point (ICP) algorithm. The candidate list used for appearance matching is dynamically generated based on the output of the surface matching component, which reduces the complexity of the appearance-based matching stage. The 3D model in the gallery is used to synthesize new appearance samples with pose and illumination variations that are used for discriminant subspace analysis. The weighted sum rule is applied to combine the two matching components. A hierarchical matching structure is designed to further improve the system performance in both accuracy and efficiency. Experimental results are given for matching a database of 100 3D face models with 598 2.5D independent test scans acquired in different pose and lighting conditions, and with some smiling expression. The results show the feasibility of the proposed matching scheme.
• Multimodal Authentication using Online Signature and Voice Steve Krawczyk, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering Regulations imposed by the HIPAA standards and the widespread deployment of wireless access points has induced the need for user authentication for accessing medical records. Biometric authentication is an obvious choice; it relies on physiological and behavioral characteristics and cannot be lost or forgotten. Our approach to this problem is to design a multi-biometric authentication system using signature and speech modalities. This combination will ensure sufficient security and user convenience.
• Automatic Extraction and Integration of Soft Biometric Traits for User Recognition Unsang Park, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering Soft-biometric features, such as gender, ethnicity, height, weight and eye color can be used in addition to the primary biometric features (face, fingerprint, iris, etc.) for personal identification. We developed a prototype system for automatic soft-biometric feature extraction. The system extracts height, face and eye color information as a user approaches a biometric sensor. The effect of the soft-biometric features is evaluated using real primary biometric data of 263 users and synthetic soft-biometric features. The results show that there are significant performance improvements by utilizing soft-biometric features.
• Dental Biometrics: Matching Dental Radiographs for Human Identification Hong Chen, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering Dental biometrics automatically analyzes dental X-ray images to identify deceased individuals, for whom other biometric features (e.g., fingerprint, face and iris) are not available. Radiographs acquired after a person is deceased are called Post-mortem (PM) radiographs, and the radiographs acquired while the person is alive are called Ante-mortem (AM) radiographs. The goal is to assign correct identities to the PM images by matching the PM radiographs against the database of AM radiographs.
• Image Enhancement and Quality Assessment for Multi-Biometric Systems Yi Chen, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering Image quality is one of the crucial factors that affect the performances of every Automated Biometric Identification Systems. A good enhancement algorithm should be adaptive to different kinds of image quality, remove noise and preserve the most representative features. But more importantly, a reliable enhancement algorithm can also assist us to evaluate the quality of an image and hence assign confidence to the features extracted in a later stage. In this work, we show how image enhancement techniques can be used for quality assessment and improve the matching performance. We also show how it can be used to help the fusion of multi-biometric features.
• Using Biometric Authentication in E-Commerce Transactions Zoo-Hyun Chae, Ying-Ju Lai, and Nala Kok-Srey, Telecommunication, Information Studies and Media Advisors: Dr. Charles Steinfield, Telecommunication, Information Studies and Media Dr. Dan Kim, Telecommunication, Information Studies and Media Identity theft is one of the fastest growing crimes of the new millennium. A technological approach for a solution against these identity crimes is the implementation of biometric authentication method in credit card transactions. Among various biometrics, one of the most prevalent and easy-to-implement authentication techniques is fingerprint scanning. This study explicitly investigates the effectiveness and applicability of biometrics in social contexts. According to the Banking Industry Technology Secretariat (BITS), a large portion of identity thefts has been occurring in online transactions. Thus we specifically examine the effects of employing biometric authentication in online settings. Also, since the current online credit card payment methods require consumers to either (1) type in the account numbers and expiration date, or (2) type in an additional personal identification number (PIN) as a protective manner, we compare these processes with (3) the biometric authentication method of fingerprint scanning.
• Robust Fingerprint Matching Karthik Nandakumar, Computer Science and Engineering Advisor: Dr. Anil K. Jain, Computer Science and Engineering Different fingerprint recognition algorithms represent the same fingerprint image in different feature spaces. Hence, these algorithms can complement each other and utilizing such complementary information can lead to an improvement in the recognition performance. In this work, we have studied the performance of three fingerprint matchers, namely, 2-D dynamic programming based minutiae matcher, ridge feature map (fingercode) based matcher, and local correlation-based matcher. Although, these matchers have different representations of the fingerprint image, they follow a similar procedure for aligning the template and query fingerprint images. We show that combining these matchers can lead to better accuracy and robustness than any of the individual matchers. We also analyze the cases where the matcher fails and put forth suggestions for avoiding such failures.
• Ensuring Security and Privacy in a Telemedicine Videoconferencing Research Project Michael Mackert, Telecommunication, Information Studies and Media Advisor: Dr. Pamela Whitten, Telecommunication, Information Studies and Media MSU researchers from the Department of Telecommunication have partnered with four rural long-term care facilities in Michigan to begin a program to provide an innovative telemedicine consultation program. Specifically, these rural nursing homes will be able to provide clinical consultations directly to the bedsides of their residents via a wireless network in each facility that connects the nursing homes to a videoconferencing network. The provision of such services has required care to be taken regarding wireless security in the nursing homes themselves, and this is only heightened by concerns regarding HIPAA and resident privacy. A description of the videoconferencing network, the wireless networks in each facility, and other security-oriented practices that have been enacted for this research project are discussed.
• Modeling the Spread of Worms over VANET Ali Khayam, Electrical and Computer Engineering Advisor: Dr. Hayder Radha, Electrical and Computer Engineering This work investigates the parameters governing the spread of active worms over VANET. To this end, we first define the average degree of a VANET node using freeway traffic parameters. The spread of a worm in congested and low-density traffic scenarios is modeled using a stochastic model of infectious disease. Analysis is provided for preemptive and interactive patching scenarios. • Computer Worm Protection in Hardware (Secure Bit*: Hardware Solution Against Buffer Overflow) Krerk Piromsopa, Computer Science and Engineering Advisor: Dr. Richard Enbody , Computer Science and Engineering We propose a new, minimalist, architectural approach, secure bit, to protect against buffer overflow and function-pointer attacks. Secure Bit is almost completely transparent to software, and has little run-time performance penalty. The goal of Secure Bit is to provide hardware support to protect against current and future generations of buffer-overflow attacks by protecting the integrity of addresses. Included is a reference to our proof that validates the mechanism of the Secure Bit. Robustness and transparency are demonstrated by emulating the hardware, and booting Linux on the emulator and running application software.
• Mining and Modeling Physiological Data Liu Yang, Computer Science and Engineering Advisor: Dr. Rong Jin, Computer Science and Engineering Wearable, unobtrusive devices that continuously monitor the vital signs of a person are becoming increasingly common, especially for home care, telemedicine, weight management as well as fitness management applications. The massive amount of physiological data generated by these devices present a great opportunity for applying data mining and machine learning techniques for modeling and extracting useful information about a person's well-being. In this poster, we report our preliminary results in applying a boosting-based method for classifying human activities from such data and an association analysis method for building summary rules describing certain activities of a person.
• A Cryptographic Watermarking Technique Huahui Wang, Electrical and Computer Engineering Advisors: Dr. Jian Ren and Dr. Tongtong Li, Electrical and Computer Engineering A new cryptographic watermarking technique is proposed in this paper. The proposed method is very simply and efficient. It can be applied to digital image watermarking in both spatial domain and spread spectrum domain. It can also be applied to audio watermarking. Our security analysis and performance analysis demonstrate that the proposed watermark method provides a good transparency, robustness and security.
• Text Mining in MedLine Sam Shaojun Zhao, Computer Science and Engineering Advisor: Dr. Joyce Chai, Computer Science and Engineering As the repository of medical documents continues to grow rapidly, automatic knowledge discovery from medical texts becomes increasingly important. One crucial step in this process is to automatically identify special biomedical names such as DNA, RNA, protein, cell type, and cell line names. Since most biomedical names have many different variations and new names can potentially be introduced overtime, the use of a medical dictionary would be insufficient. This poster presents a system that automatically identifies biomedical names from 1,381,132 abstracts in the MedLine database using natural language processing techniques. Based on the recognized names, our next step is to discover relations between those entities (e.g., gene relations).
Working Lunch The Michigan State University Initiative: Synergistic Research Session Moderator: Presenter: |