Students Develop Real-World Web and Pervasive Computing Systems

Students Develop Real-World Web and Pervasive Computing Systems

Emerging Computer Applications to Multidisciplinary Security Issues Charles Tappert and Sung-Hyuk Cha School of Computer Science and Information Systems Previous Research Experience Charles Tappert 26 years research at IBM

7 years teaching at West Point Speech recognition and processing Handwriting recognition and pen computing Research on handheld/wearable computers Sung Cha 3 years graduate work at the world renown Center for Excellence in Document Analysis and Recognition (CEDAR), SUNY Buffalo

Individuality of handwriting 2 years research at Samsung Medical information systems Our Current Areas of Research Related to Security

Handwriting and Forensic Document Analysis Speech/Voice Related Studies Individuality of Handwriting, Voice, Iris (fundamental studies for biometric authentication) Related Pattern Recognition Research Wearable/Mobile/Pervasive Computing Research Forensics Applications Security Related Research/Projects

D.P.S. Dissertations M.S. Dissertations Graduate and Undergraduate Students Projects CS615-616 Software Engineering CS631 Computer Vision CS632 Pervasive Computing Research Seminar CS396 Pattern Recognition

Examples of Security-Related Research Studies Security-Related Research Publications NSF Funding Proposals Security Related D.P.S. Dissertations An Efficient First Pass of a Two-Stage Approach

for Automatic Language Identification of Telephone Speech, Jonathan Law (2002) Information Assurance Strategic Planning: A Taxonomy, Steven Parshley (2004) A Cybercrime Taxonomy, Vincent Gisonti (2004) Real-time Trifocal Vision with Locate Positioning System, Yi Rong (2004) Stego-Marking in TCP/IP Packets, Eric Cole (2004) The Computer Forensics and Cybersecurity Governance Model, Kenneth Brancik (2005) Security Related M.S. Dissertations Forged Handwriting Detection, Hung-Chun Chen (spring 2003)

Speaker Individuality, Naresh Trilok (fall 2003) More coming Security Related Projects

Handwriting Forgery Detection, Forgery Quiz System Recognizing a Handwriters Style/Nationality Emergency Pre-Hospital Care Communication System Eigenface Recognition System Interactive Visual Systems (collab. with RPI, NSF funding?) Object Tracking System (Surveillance) Object Segmentation (X-ray scan) Biometric Authentication (Fingerprint, Iris, Handwriting, Voice) Others: Steganography, Wireless Security, Forensics, Spam Detection, Language Classification from Text Project Customers/Sources

Pace University School of Computer Science and Information Systems Dyson College of Arts and Sciences Lubin School of Business Lienhard School of Nursing Department of Information Technology Doctor of Professional Studies in Computing Program Office of Planning, Assessment, Research, and

Academic Support Outside Organizations Northern Westchester Hospital Columbia Presbyterian Medical Center Psychology Department at SUNY New Paltz Yonsei University, Korea CEDAR, SUNY Buffalo Rensselaer Polytechnic Institute IBM T.J. Watson Research Center

Benefits of Student Projects Stellar real-world learning experience for students Customers receive valuable systems Promotes interdisciplinary collaboration and Pace and local community involvement

Furthers student and faculty research Enhances relationships between the university and local technology companies Increases national recognition of the university Examples of Security-Related Research Studies Forgery Detection Interactive Visual System Speaker Individuality Forgery Detection: Key Idea

Forensic literature indicates that successful forgers often forge handwriting shape and size by carefully copying or tracing the authentic handwriting Exploit computing technology to investigate this and possibly to develop techniques to aid forensic document examiners Forgery Detection: Hypotheses

Good forgeries those that retain the shape and size of authentic writing tend to be written more slowly (carefully) than authentic writing Good forgeries are likely to be wrinklier (less smooth) than authentic handwriting Forgery Detection: Methodology Sample collection: online, scan to get offline Feature extraction: Speed, Wrinkliness Statistical analysis

(a) (a) Number of (b) in the boundary = 69 (b)(b) Number Number of of in in thethe boundary boundary= =3232 Fractal Measure of Wrinkliness boundary _ in _ high _ res. / log(2)

Wrinklines s log boundary _ in _ low _ res. 69 Wrinklines s log( ) / log(2) 1.1085 32 Forgery Detection: Experiment 10 subjects, each wrote

3 authentic handwriting samples 3 forgeries of each of the other 9 subjects 30 authentic and 270 forged samples Significance results (T-test) Forgeries are written slower: p = 5.90E-09 Forgeries are wrinklier: p = 0.0205 Interactive Visual System (IVS) IVS is a technology, not just a flower identification application We also have

preliminary results on flag recognition, and we plan to explore the applications of sign, IVS Motivation Image recognition can be a difficult problem Modern AI and pattern recognition techniques try to automate the process that is, they do not include the human in the equation Humans and computers have different strengths

Computers excel at large memory and computation Humans excel at segmentation We propose combining human and computer to increase the speed and accuracy of recognition IVS Flower User Interface

Load Flower Image Select Features Identify Previous 3 Hits Next 3 Hits Store New Flower Auto Feature Extract List Extracted Features IVS Flower Shape Model IVS: Flag Recognition

We have extended the Interactive Visual System to other applications, and have preliminary results on flag recognition Demonstration by Dr. Sung Cha IVS: NSF Proposal Applications Foreign Sign Recognition Face Recognition

Shape model: rectangle Shape model: 3D face template Skin Lesion Recognition Shape model? Speaker Individuality Hypothesis: a persons voice is unique

and therefore we can verify the identity of an individual from his/her voice samples Methodology: use a statistically inferable dichotomy (verification) model that Dr. Cha has used to show handwriting individuality Speaker Individuality: Methodology

Segment common portion of utterance: My name is Compute spectral data: output from 13 filters every 10 msec Extract fixed number of features per utterance from the spectral data Use the dichotomy (verification) model to obtain experimental results Speaker Individuality: Segmentation My name is from Two Speakers Neural Network Dichotomy Model ( f1x , f1 y )

( f1x , f 2x ,..., f dx ) Feature Extraction ( f 2x , f 2y ) Distance computation Same/ Different ( f1 y , f 2y ,..., f dy ) ( f dx , f dy ) Speaker Individuality:

Experiments 10 samples from each of 10 speakers 450 intra-speaker distances 4500 inter-speaker distances Train NN on a subset of the intraspeaker and inter-speaker distances Test on different subsets 94 percent accuracy 98 percent with bad samples removed

Security-Related Research Publications http://csis.pace.edu/csis/cgi-front/sec/s ecurity.pl?cat=11 Security Related Funding Proposals NSF 01-100, CISE-HCI

Interactive Visual Processing Collaboration with RPI Submitted January 8, 2004 NSF 03-602 Computer Vision Individuality Studies (fundamental studies for biometric authentication) Submitted December 19, 2003

Recently Viewed Presentations

  • Estimating river discharge from the Surface Water and Ocean ...

    Estimating river discharge from the Surface Water and Ocean ...

    Their approach draws from Haddeland et al. (2006) and Hanasaki et al. (2006). They represent reservoir storage in 7000 reservoirs globally with surface area larger than 0.1 km2 (GRAND database). We intend to adapt the Biemans et al. method to...
  • PowerPoint 프레젠테이션 - Daum

    PowerPoint 프레젠테이션 - Daum

    Author: kim chw Title: PowerPoint 프레젠테이션 Last modified by: kim chw
  • Test Taking Strategies - Little Rock School District

    Test Taking Strategies - Little Rock School District

    Answers the easiest questions first, but be sure to go back to those questions you skipped. Multiple Choice Questions Do not change your answers unless you are very uncertain about your first answer choice. ... Double-check the test number in...
  • PRESENCE Bonus: Biblical Foundatons Part 3 Secular Objections

    PRESENCE Bonus: Biblical Foundatons Part 3 Secular Objections

    How do you understand the arc of the bible story from Genesis to Revelation as the story of the Eucharist? ... About six-in-ten (63%) of the most observant Catholics — those who attend Mass at least once a week —...
  • BBC - Science with Mr. Louie

    BBC - Science with Mr. Louie

    Billiard Ball Model (1803) - John Dalton viewed the atom as a small solid sphere.-Dalton developed the first atomic theory model of the atom Atom Model History - Each element was composed of the same kind of atoms. - Each...
  • Aquatic Biodiversity - Zamora's Science Zone

    Aquatic Biodiversity - Zamora's Science Zone

    < 10% of world's ocean area but contains 90% of marine species. Most commercial fisheries. Most ecosystems—estuaries, coastal wetlands, mangrove forests, and coral reefs—have high NPP. ample supplies of sunlight and plant nutrients, which come from land.
  • Warm-Up: Take a  sheet. Read the following two

    Warm-Up: Take a sheet. Read the following two

    Read Aloud Trino's Choice (Chapter 13 summary and Chapter 14) (make connections to prior experience andanalyze character development) 10. Classroom Conversation. 10. Whole Group (Analyze the Development of a Theme) 35. Work Time (2.15 Analyze a Theme) 5. Wrap-Up
  • Xen Bart Miller Outline Definition and goals Paravirtualization

    Xen Bart Miller Outline Definition and goals Paravirtualization

    CPU. Protection. x86 has 4 privilege levels, known as Rings. Ring O is highest and Ring 3 lowest privilege. For Xen, the VMM executes in Ring O, the Guest OS executes in Ring 1, and the user programs execute in...