Transcript Networks - cse services - The University of Texas at Arlington

Faculty Research Areas
Labs/Centers
Meetings
Fall 2010
1
Areas
 Artificial Intelligence
 Bio-Informatics
 Databases
 Graphics, Image Processing and Multimedia
 Networks
 Pervasive Computing
 Software Engineering
 Systems and Architecture
 Security
Fall 2010
2
Artificial Intelligence
Manfred Huber
Farhad Kamangar
Vassilis Athitsos
Gian Luca Mariottini
Fall 2010
3
Manfred Huber
Research Projects
• Personal Service Robots
• Hierarchical Skill Acquisition
• CONNECT - Information Technologies
for the Disabled
Contact:
[email protected]
(GACB114)
Fall 2010
4
Farhad Kamangar
Research Projects
• Computer Vision
• Neural Networks
• Robotics
• CONNECT - Information Technologies
for the Disabled
Contact:
[email protected] (GACB 112)
Fall 2010
5
Bio-Informatics
Dr. Fillia Makedon
Dr. Heng Huang
Dr. Chris Ding
Dr. Jean Gao
338 Nedderman Hall
Phone: (817) 272-3628
E-mail: [email protected]
URL: http://crystal.uta.edu/~gao
Dr. Nikola Stojanovic
301 Nedderman Hall
Phone: (817) 272-7627
E-mail: [email protected]
URL: http://ranger.uta.edu/~nick
Fall 2010
6
Fall 2010
http://www.washbac.org/images/farside.gif
7
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
Fall 2010
8
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
it will probably happen exactly that way
Fall 2010
9
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
it will probably happen exactly that way
 Modern high-throughput technologies are generating
tremendous volume of data - somebody needs to store and
manipulate the data, generate reports and share them with
the scientific community.
Fall 2010
10
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
it will
probably happen
exactly that way
 Modern high-throughput
technologies
are generating
tremendous volume of data - somebody needs to store and
manipulate the data, generate reports and share them with
the scientific community.
Fall 2010
11
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
it will probably happen exactly that way
 Modern high-throughput technologies are generating
tremendous volume of data - somebody needs to store and
manipulate the data, generate reports and share them with
the scientific community.
 Can we turn that data into information, and eventually
knowledge?
Fall 2010
12
What is BIOINFORMATICS?
 Have you ever thought that a cure for cancers could be
developed by people working at their computers?
it will probably happen exactly that way
 Modern high-throughput technologies are generating
tremendous volume of data - somebody needs to store and
manipulate the data, generate reports and share them with
the scientific community.
 Can we turn that data into information, and eventually
knowledge?
Fall 2010
13
http://bioinformatics.ubc.ca/about/what_is_bioinformatics/
Fall 2010
14
http://bioinformatics.ubc.ca/about/what_is_bioinformatics/
Fall 2010
15
http://bioinformatics.ubc.ca/about/what_is_bioinformatics/
Fall 2010
16
Fall 2010
17
Biotechnology and pharmaceutical
industry
 Biotechnology and pharmaceutical industry revenues are
estimated at hundreds of billions of dollars annually.
 The industry's claim is that they spend $800 million on
research & development for every new drug which
receives FDA approval.
 Much of the R&D efforts are pursued computationally
these days.
Fall 2010
18
Biotechnology and pharmaceutical
industry
 Biotechnology and pharmaceutical industry revenues are
estimated at hundreds of billions of dollars annually.
 The industry's claim is that they spend $800 million on
research & development for every new drug which
receives FDA approval.
 Much of the R&D efforts are pursued computationally
these days.
 This is a large and growing industry - whether in R&D or
just software support, you may see yourself working
for one of these companies in a few years.
Fall 2010
19
http://bioinformatics.uta.edu
Fall 2010
20
Bioinformatics lab projects
 Motif discovery in DNA sequences.
 Identification and characterization of mobile elements in
DNA.
 Studying structure and conservation patterns in genomic
sequences.
 Characterization of chromosomal recombination patterns.
 Studying human genetic variation and its relation to disease
susceptibility.
Fall 2010
21
Bioinformatics lab projects
 Motif discovery in DNA sequences.
 Identification and characterization of mobile elements in
DNA.
 Studying structure and conservation patterns in genomic
sequences.
 Characterization of chromosomal recombination patterns.
 Studying human genetic variation and its relation to disease
susceptibility.
Research funded by the National Institutes of Health, and
preformed in collaboration with UTA Biology Department
and the University of Texas Southwestern Medical Center
in Dallas.
Fall 2010
22
UT Arlington
http://www.biotconf.org
Fall 2010
23
Databases
Sharma Chakravarthy
Ramez Elmasri
Leonidas Fegaras
Gautham Das
Chengkai Li
Fall 2010
24
Information Technology Laboratory
Prof. Sharma Chakravarthy
Email: [email protected], URL: http://itlab.uta.edu/sharma
Funding Sources: NSF, Spawar, Rome Lab, ONR, DARPA, TI, MCC
Select Projects
 InfoMosaic (information
integration from
heterogeneous sources)
 MavEStream: (Event
and Stream Processing)
 Active Technology
(Push Paradigm, pub/sub,
event-driven architectures)
 WebVigiL: (General
Purpose Change Monitoring
for the web)
 Mining: Graph, Text,
Assoc Rules
 Prediction of Event
Patterns
Select Publications
People
1.
PhD Students –
2.
3.
4.
5.
6.
7.
8.
 Information Search,
Filtering, and
classification
9.
 Information Security
10.
 Mobile Caching
1.
R. Adaikkalavan and S. Chakravarthy, Event
Specification and Processing for Advanced Applications:
Generalization and Formalization, DEXA Sep 2007
A. Telang, R. Mishra, and S. Chakravarthy, Ranking Issues for
Information Integration, DBrank workshop (ICDE 2007),
Turkey, 2007.
S. Savla and S. Chakravarthy, Efficient Main Memory
Algorithms for Significant Episode Discovery, To appear in
the Int’l Journal of Data warehousing and Mining, 2006.
R. Balachandran, S. Padmanabhan, S. Chakravarthy
Enhanced DB-Subdue: Supporting Subtle Aspects of
Graph Mining Using a Relational approach in PAKDD, 2006
A. Srinivasan, D. Bhatia, and S. Chakravarthy, Discovery of
Interesting episodes in Sequence Data, in 21st ACM SAC,
Data Mining Track, 2006.
M. Aery, S. Chakravarthy: eMailSift: Email Classification
Based on Structure and Content in IEEE ICDM 2005
H. Kona, S. Chakravarthy, and A. Arora, SQL-Based
Approach to Incremental Association Rule Mining, in
ADBIS Workshop on DMKD, 2005.
Q. Jiang, R. Adaikkalavan and S. Chakravarthy, NFMi: An
Inter-domain Network Fault Management System. IEEE
ICDE, 2005.
R. Adaikkalavan, and S. Chakravarthy: Active Authorization
Rules for Enforcing Role-Based Access Control and its
Extensions, PDM Workshop, IEEE ICDE, 2005.
L. Elkhalifa, R. Adaikkalavan, and S. Chakravarthy, InfoFilter:
A System for Expressive Pattern Specification and
Detection Over Text Streams, ACM SAC, 2005.
….
Mr. Aditya Telang (Adi)
Ms. Roochi Mishra
Masters Students –
Mr. Mayur Motgi
Mr. Supreet Chakravarthy
Mr. Aamir Syed
Group Meeting:
1 Pm to 2 Pm on Fridays
in NH 232
Fall 2010
25
A Distributed Middleware-Based Architecture for FaultTolerant Computing Over Distributed repositories
 Semi-joins
uav6
 Compression
 Replication
 Smart Routing
uav
2
uav4
uav3
uav1
uav5
…
Ground controller 1
Ground controller 2
Ground controller n
Fall 2010
26
Network of computing nodes:
Unmanned vehicles, Sensors, Robots, PCs ,
Servers, Ground Controlling devices
Limited Resources
Mobility
Heterogeneity
Disconnections
Queries, Tasks, Requests, Continuous
Queries Publish/Subscribe
SOA Distributed Middleware
Fault
Tolerance
Services
Task planning
Composition
Context-aware
Resource Management
Raw Data / fused
data /data from
other nodes
Query
Capability
Context/
Knowledge
Base
Join computation
pub/sub
Notification
Data management
Publish
Subscribe
Capability
Local
fusion/Materiali
zation
Fall 2010
27
Ramez Elmasri
Professor
Databases
Distributed XML Querying and Caching
Object-Oriented Databases
Keyword-based XML Query Processing
Sensor Networks
Energy-Efficient Querying of Sensor Networks
Combining RFID and Sensor Networks
Indexing of Sensor Networks Data
Bioinformatics
Modelling Complex Bioinformatics and Biomedical
Data
Mediators for Accessing Heterogeneous Data
Sources
Fall 2010
28
Leonidas Fegaras
Areas of interest:
 Databases




Associate Professor
(PhD: UMass 1993)
Web Databases and XML
Object-Oriented Databases
Query Processing and Optimization
Data Management on Peer-to-Peer Systems
 Programming Languages
 Functional Programming
 Program Optimization
Fall 2010
29
Research Review
Gautam Das
 Database Exploration
 Web/Information Retrieval searching techniques in databases
 OLAP, Data Warehouse, Approximate Query Processing
 Data Mining
 Clustering, Classification, Similarity models, Time-Series
Analysis
 Algorithms
 Graph Algorithms, Computational Geometry
More information available at
http://ranger.uta.edu/~gdas/website/research.htm
Fall 2010
30
Chengkai Li
Assistant Professor
http://ranger.uta.edu/~cli
[email protected]
 The Innovative Database and Information Systems Research (IDIR) Lab
http://idir.uta.edu , GeoScience 237
Jared Ashman, Avinash Bharadwaj, Ebrahim Cutlerywala, Sunny Hasan, Naeemul Hassan,
Angus Helm, Nandish Jayaram, Pat Jangyodsuk, Xiaonan Li, Vikramark Singh, Ning Yan
 Research Areas
Databases, Web Data Management, Information Retrieval, Data Mining
 Specific Topics

Data Retrieval and Exploration, Ranking and Top-k Queries; Web
Search/Mining/Integration, Web Databases, Query Processing and Optimization,
OLAP and Data Warehousing, Cloud Computing, Database Testing, XML
 Projects: Search the Database and Query the Web

Computational Journalism

DBTest: Database Application Testing

Entity-Centric Enterprise Information Management

BestCloud: Query Optimization for Cloud Computing

RankSQL: Ranking and Top-k Queries, Database Exploration

SetQuery: Set-Oriented OLAP Queries

WebEQ: Querying and Exploring Structured Information on the Web
31
Two Demos from WebEQ project
 Facetedpedia
http://idir.uta.edu/facetedpedia/
 Entity-Relationship Queries
http://idir.uta.edu/erq/
32
Graphics Image Proc., Multimedia
Ishfaq Ahmad
Multimedia Authoring, Compression,
Communication
Video Processing,
Next Generation TV
Network Security
Parallel Algorithms
Dr. Gutemberg Guerra-Filho
Computer Vision, Animation, and
Humanoid Robotics
Fall 2010
33
Prof. Ishfaq Ahmad
 Dr. Ahmad works closely with federal
agencies, Arlington police and multimedia
industry.
 Several projects in power-aware video
compression, multimedia systems, next
generation TV are being pursued in his lab.
Fall 2010
34
High-Performance
Ishfaq Ahmad
 Resources Management in Parallel
and Distributed Systems
Power Management in Data Center
and Distributed Systems
Fall 2010
35
Institute for Research in
Security (IRIS)
Ishfaq Ahmad
A Multi-disciplinary center focusing on
infrastructure, people, and environmental security
http://www.iris.uta.edu/
Fall 2010
36
Networks
Sajal Das
Mohan Kumar
Gergley Zaruba
Hao Che
Yonghe Liu
Fall 2010
37
Sajal K. Das
Center for Research in Wireless Mobility
and Networking (CReWMaN)
Sajal K. Das, Mohan Kumar
Yonghe Liu, Hao Che
[email protected]
URL: http://crewman.uta.edu
Woolf Hall 411,413,
Tel: 2-7409
[Networking, Mobile Computing and Parallel Computing Research Group]
Fall 2010
38
Mohan Kumar
Pervasive and Mobile Computing
Sensor Systems

Pervasive Computing






Uniform Information Access in




Distributed, mobile and pervasive systems
Caching, prefetching, and broadcasting
Data management
Peer-to-Peer (P2P) Systems




Middleware
Service creation, composition and deployment
Prototype development
Sensor networks and smart environments
Information Fusion in pervasive/sensor environments
Information and service sharing
Efficient communication and collaboration
Security and privacy
Recommended courses before
starting thesis work:
CSE5311, CSE5346,CSE5306 and
CSE5347/5355
Directed Study
Active and Overlay Networking


Novel protocols
Role in mobile, pervasive and P2P computing
Fall 2010
39
Gergely Zaruba
Research Projects
 Personal Area Networks
 Heterogeneous Wireless Networks
 Architecture, Admission Control and Handoff
 Optical Networks
 Optical Burst Switching, Routing, QoS Provisioning
 Traffic Modelling
Contact:
[email protected]
(GACB 112)
Fall 2010
40
Hao Che
 Embedded hardware/software design for NG
network processors
 Traffic engineering
 Implementation issues and software
development
 MPLS path protection and fast rerouting
 Routing redundancy
 Traffic modeling for wireless networks
Contact:
http://crystal.uta.edu/~hche/
[email protected]
Fall 2010
41
Yonghe Liu
 Sensor network and security
 Prototyping and experimental study
 Theoretic design and analysis
 Cross layer optimization
 Channel dependent performance
 Software security
 Design and analysis
In need of
 Strong mathematic skill (probability/signal processing/number
theory/etc), or
 Strong programming skill (hardware/software)
Contact:
http://ranger.uta.edu/~yonghe/
Fall 2010
42
Software Engineering
David Kung
Yu Lei
Dr. Christoph Csallner
David Levine
Fall 2010
43
David Kung
 Agent-Oriented Software Engineering
 Testing Object-Oriented Software
 Expert System for Design Patterns
 Formal Methods for Quality Assurance
 Fault Tolerance and Automatic Recovery
Using Dynamic Class Diversity
Contact:
http://ranger.uta.edu/~kung/kung.html
Fall 2010
44
Yu Lei
 Concurrent and real-time software systems
 Race analysis, Deterministic Execution
Environment, Reachability Testing, State
Exploration-Based Verification
 Automated software testing
 Object-Oriented Testing, Component-Based
Testing, Combinatorial Testing
Contact:
http://ranger.uta.edu/~ylei
Fall 2010
45
David Levine
High Throughput Computational
Science: Clusters and Grids::
David Levine, CSE@UTA
Projects: (Computers applied to:)
High Energy Physics, Bioinformatics,
Medical Informatics, People with
Disabilities, Streaming Processing, other..
Fall 2010
46
Check out the
lab: NH 246
Software Engineering
Research Center
Faculty members:
Dr. Christoph Csallner
Dr. Dave Kung
Dr. Jeff Lei
Fall 2010
47
Fall 2010
48
Software Engineering
 Software has become
pervasive in modern society
 Directly contributes to quality of life
 Malfunctions cost billions of dollars every year,
and have severe consequences in a safe-critical
environment
 All about building quality software,
especially for large-scale development
 Requirements, design, coding, testing,
maintenance, configuration, documentation,
deployment, and etc.
Fall 2010
49
THE Best Job in America
What is the 2nd best job?
Go for a PhD in Software
Engineering!!
Fall 2010
50
Great Impact
Fall 2010
51
Quotes from Dr. Parnas
Extracted from his ACM Fellow Profile
http://www.sigsoft.org/SEN/parnas.html
Fall 2010
52
Current Research Projects
 Object-Oriented Software Analysis and Testing





(Dr. Kung)
Software Security Analysis and Testing (with Drs.
Kung and Liu)
Pervasive Context-Aware Computing (with Dr.
Kumar)
Formal Testing and Verification of Concurrent
Software Systems (with GMU)
Automated Combinatorial Testing for Software
(with National Institute of Standards and
Technology)
Interaction Testing of Web Applications (with
Fall 2010
UMBC)
53
Current Research Projects
 Hybrid static-dynamic program analyses
 Automatic test case generators
 JCrasher, Check ‘n’ Crash, DSD-Crasher
 New: Testing of database-centric
applications
 OrmCheck with
 ToDo: Support complex languages like UML
 New: Dynamic symbolic invariant detector
 Pex/DySy with
 ToDo: Scale analysis to large applications
 ToDo: Add static knowledge to dynamic
Fall 2010
54
Fall 2010
55
If you want to improve..
..come talk to
us
Fall 2010
56
Embedded Systems :: Roger Walker
Embedded Systems for Transportation
Applications:
 Real-time Multi-core Systems for Embedded
Applications
 Stochastic Modeling From Sensor Measurements
 Development of Special Measurement Systems
for Transportation Related Applications
Contact:
http://ranger.uta.edu/~walker/
Fall 2010
57
Design and Development of a Mobile Bridge
Design
and Development of System
a Mobile
Monitoring/Measurement
Bridge Monitoring/Measurement
System
Profiler
Gyroscope
Surface
Data
Integrate Data
Scanning
Laser
Video
Video
Video
Surface
Structure
Data
Data
Design and Development of Portable
Real-Time Embedded
Measure and Control Systems
Current Research Projects supported by Texas
Department of Transportation, Federal Highway
Administration, & Intel
Information Security
Donggang Liu
Matt Wright
Fall 2007
60
Jobs in Infosec
Fall 2007
61
One aspect of security
Operational Security
 Classified material can be leaked based
on how it’s used or through side effects
Domino’s Pizza Anyone?
Last Wednesday, he adds, "we got a lot of orders, starting around midnight. We figured
something was up." This time the news arrived quickly: Iraq's surprise invasion of
Kuwait.
"And Bomb the Anchovies", Time, p. 13, 8/13/90
Fall 2007
62
Border Security with WSNs
PIs: Donggang Liu, Sajal K. Das, Matthew Wright
Post-Doc: Jun-Won Ho
Students: Andy Fox, Na Li, Nabila Rahman, Mayank Raj, Kartik Siddhabathula
Funded in part by the National Science Foundation
 Goal
 Intruder tracking
 Intruders




Corrupt many sensors
Jam wireless channels
Destroy key infrastructure
Seek gaps in the sensing
coverage
Website:
http://isec.uta.edu/borde
Fall 2007
63
Wireless and System Security ::
Donggang Liu
Security in wireless sensor networks
 key management, security of services such as localization, routing,
clustering etc.
Integrity of wireless embedded devices
 Code integrity, tamper-resistant techniques
Software and system security
 Security testing, detection of malicious code
Contact:
http://ranger.uta.edu/~dliu
Fall 2007
64
Matthew Wright
Internet Privacy
Robust P2P
Distributed Twitter
Sensors/Mobile/Social/Ubicomp
…
Contact: http://isec.uta.edu/mwright
Fall 2007
65
Assist Laboratory
F. Kamangar, M. Huber, D. Levine, G. Zaruba
Computer Science and Engineering Department
The University of Texas at Arlington
Fall 2010
66
Information Technologies for Persons
with Disabilities and Health Care
• Assistance for Persons with
Disabilities
• Communication devices and
technologies
• Intelligent assistive devices
• IT for improved care
• Information Technologies for
Healthcare and Aging
• Automatic health monitoring
• Intelligent environments
• IT to improve uniform
communication needs
Fall 2010
67
Connect - Intelligent Communication
Technologies for Disability & Health Care
• Intelligent communication
services connect individuals
with care providers and with
important information
• Seamlessly connected devices
• Adaptive interfaces
• Universal underlying
software architecture
• Intelligent information
analysis and interpretation
• Seamless, omnipresent
access to information
Wireless Communication
Provider
Servers, Databases, Web pages
Internet
Technical support
Human Service Providers
Clients
Fall 2010
68
Assistive Technologies
• Computer Technologies Can
Enhance Assistive Devices
• Ayuda – Intelligent wheelchair
• Autonomous navigation
capabilities
• Environment sensing
• Integration of computer control
and user instructions
• Force feedback technologies to
enhance interaction capabilities
for persons with physical
disabilities
Fall 2010
69
Health Monitoring and Intelligent
Environments for Aging in Place
• Wirelessly Connected Sensors
Provide Health Information and
can Improve Quality of Life
• Health sensors can monitor conditions
and detect problems
• Wireless communications permit
continuous monitoring
• Prediction and modeling technologies
facilitate automatic analysis of the data
• Communication technologies allow
connectivity to physician
• Sensors in the environment allow
automation of important functions and
assistance
• Monitoring and assistance for Aging in
Place
Fall 2010
70
AI and Robotics Laboratory
M. Huber, F. Kamangar
Computer Science and Engineering Department
The University of Texas at Arlington
Fall 2010
71
Adaptation and Learning in
Robots and Computer Systems
• Personal Service Robots
• Service robots have to interact with
people
• Programmability by unskilled users
• Robustness in real world situations
• Variable Autonomy
• Robots have to be easy to program
• Robots should understand any kind
of user command
• Cognitive Development
• Computer systems have to learn how
to act and reason in the world
Fall 2010
72
Robot Imitation – Programming by
Demonstration
• Learning to Sense
• Imitating robots have to be able to
interpret their observations
• Learning to Relate Human
Demonstrations to Robot
Actions
• Learning to extract the important
aspects of human actions
• Translating human actions into
corresponding robot controls
• Learning to Interpret Task
Requirements
• Robots have to be able to learn to
ignore dangerous commands
Fall 2010
73
Hierarchical Skill Learning /
Cognitive Development
• Learning Behavioral
Strategies
• Adaptation to unknown
conditions
• Automatic extraction of
subtasks
• Hierarchical Learning
• Learning with abstract actions
• Learning using state
abstractions
• Facilitation of incrementally
more complex behavior
Fall 2010
74
Robot Activities and Platforms
• Robot Soccer (RoboCup)
• Autonomous robotic soccer with
robot dogs
• Student team
• Computer Game Trials
• UCT – Urban Combat Testbed
Fall 2010
75
The HERACLEIA Human Centered Computing Lab
Vicon Camera
Vicon Motion
Capture
System
HERACLEIA was a thriving outpost of Hellenic culture
south of the Black Sea. Symbolizes a world where
technologies are placed at the service of humans, esp.
those needing special help, and bringing out the
human side of technology.
Bioloid Robot
Peoplebot
SunSPOT
Wireless
Sensor Node
Fall 2010
76
The Heracleians
Fillia Makedon (Director)
Professor Chair of Computer Science and Engineering
Current work:
Computational Multimedia Applications, Multimedia Authoring and
Retrieval, Analysis of fMRI Brain Activations, and Electronic Commerce
Zhengyi Le (Assistant Director)
Research Assistant Professor
Current work: Security, Privacy, and Collaboration System
Kyungseo Park
Academic Interests:
Data Mining in Wireless Sensor Networks
Fall 2010
77
Some of our Security Work
 Mobile Device Protection against Loss and Capture (PETRA09)
 Our forward secure two-party signature scheme provides stronger device
authentication to make it work against impersonation
 Privacy-Enhanced Opportunistic Networks (PSPAE09)
 group mobile nodes together to randomly detour the traffic to protect from
timing traffic analysis (which leads to privacy leakage)
 Providing Location Privacy (PETRA08)
 use dynamic zone to mix some location records of some moving objects to
protect against tracking
 Source Location Privacy (SecureCom08)
 hide event messages into maintenance messages so that an attacker can not track where an
event is happening (if source location information is sensitive)
 Preventing Unofficial Information Propagation (ICICS07)
 use short-lived certificates with forward secure signatures to make the information on a
certificate not verifiable shortly after usage
 Challenges
 how to apply expensive (resource consuming) cryptosystems in mobile,
portable, assistive devices (computationally limited)
 faster encryption methods that a light mobile device can afford.
 anti-data-mining mechanisms and privacy preserving technologies to address
the increasing public concerns on privacy information leakage.
Fall 2010
78
Data Sharing: Open Collaboration
Support: Group, Role, File Sharing, Recommendation
Files and
access
policies
Groups, Roles, Files
Roles and
Required
attributes
Top 10
recommendations
Recommendations
Group name,
Description and
Expiration date
Group Operations
Fall 2010
79
Behavioral Markers:
Making Genotype-Phenotype Correlations
 Certain genetic anomalies lead to certain diseases/disabilities (phenotype is
any demonstration of the conditions, such as a scan).
 Understanding Genotype-Phenotype correlations may help create more
effective treatments.
 Challenges:
 How to correlate certain medical conditions with
observable behaviors or physiological conditions.
 How to use correlations to enhance decision making.
 How to analyze the effects of medical
treatments and adapt to patient
condition.
Deletion 9q34.3 syndrome
Fall 2010
80
80
@Home Apartment
Fall 2010
81
Active Service Robots
Problem: When abnormal event occurs,
how can a robot decide what to do?
Approach:
• robot investigates and prompts human to respond by keyboard, touch
screen, or voice.
• Human cancels/confirms alarm or no action.
• Then robot makes a decision based on the available streams of sensor
and human information using partial order Markov decision processes.
Challenges:
• Setting up the hierarchy of decision making to determine what level of
action is appropriate by funneling the events of four different data streams
into the partial order Markov decision process.
• Able to access additional sensors to confirm the status of the human
• Evaluating and testing the correctness of the decisions.
Yong Lin, Eric Becker, Kyungseo Park, Zhengyi Le, Fillia Makedon Decision Making in Assistive Environments using
Multimodal Observations Proceedings of the 2nd International Conference on Pervasive Technologies Related to
Assistive Environments (PETRA'09), Corfu, Greece, June 9-13, 2010.
Fall 2010
82
Conference Proceedings: ACM will be the publisher of the proceedings of the PETRA
conference
Selected papers will be in invited to the International Journal of Functional Informatics
and Personalized Medicine, eJeta, and Journal of Personal and Ubiquitous Computing
WWW.PETRAE.ORG
PETRA 2010
Fall 2010
83
Research at the
Vision-Learning-Mining Lab
Vassilis Athitsos
University of Texas at Arlington
Fall 2010
84
American Sign Language
 0.5-2 million users in the US.
 Complete and independent language.
 Not a signed version of English.
Fall 2010
85
Looking Up
a Sign
 It is easy to go
from an English
word to ASL.
Fall 2010
86
Looking Up
a Sign
 It is easy to go
from an English
word to ASL.
 It is hard to
look up the
meaning of a
sign.
Fall 2010
87
Looking Up a Sign
 Our goal: automated
sign lookup.
 Input: video of a sign.
 The user performs the
sign in front of a camera.
 Output: best matches in
a database of 3000
signs.
Fall 2010
88
Research Directions
 Challenging problems in vision, learning,
database indexing.
 Large-scale motion-based video retrieval.
Need for developing novel atabase indexing methods
 Efficient large-scale multiclass recognition.
How can a computer learn to recognize 3000 signs?
 Learning complex patterns from few examples.
Fall 2010
89
Object Detection
Fall 2010
90
Object Detection
Fall 2010
91
Parsing Satellite Images
 Research goals:
 Accuracy.
 Efficiency.
Fall 2010
92