Transcript Artificial Intelligence

Copyright R. Weber
INFO 629 Concepts in Artificial
Intelligence
Fall 2004
Professor: Dr. Rosina Weber
Copyright R. Weber
My introduction
• Assistant Professor, Information Science & Technology,
Drexel University
• Navy Center for Applied Research in Artificial Intelligence,
Naval Research Lab
• Doctoral degree from Production Engineering Program
(UFSC, SC/BRAZIL + USF, FL/USA)
• Master’s degree in Artificial Intelligence & Operations
Research
• Bachelor’s Business Administration
• Industry experience
• Solving knowledge management problems with CI/AI
methods, particularly CBR
• Publications at
http://www.pages.drexel.edu/~rw37/publications.html
Copyright R. Weber
INFO 629 topics
1.
2.
3.
4.
5.
6.
7.
Expert Systems
Intelligent Tutoring Systems
Case-based reasoning
Search
Machine Learning, Data Mining
Neural Networks, Genetic Algorithms
Natural Language Processing
Copyright R. Weber
What is AI?
Why do we need AI ?
Copyright R. Weber
Introduction to AI
•
•
•
•
definition of AI
AI concepts
AI tasks
AI applications
Copyright R. Weber
What is AI?
Copyright R. Weber
What’s your definition of AI?
Copyright R. Weber
What is AI (from R&N)
Copyright R. Weber
What is AI (from R&N)
Copyright R. Weber
What is AI (from R&N)
Copyright R. Weber
What is AI (from R&N)
Copyright R. Weber
What is AI (from R&N)
Copyright R. Weber
Artificial Intelligence
Artificial Intelligence is the study of
computational models to perform
tasks normally associated with
rational behavior manifested as
reasoning, perception, and
appropriate actions and reactions.
Copyright R. Weber
Artificial Intelligence
Artificial Intelligence is the field of
study dedicated to the study and
design of computational models
that perform tasks that are typically
considered “human”. These tasks
may entail use of knowledge,
reasoning, or physical abilities.
Copyright R. Weber
Artificial Intelligence
•Study and design of computational
models (purposes, methods)
– Study, solve problems e.g. assisting,
replacing
– Methods use techniques that are new or
adapted from other fields
•Perform tasks
– What are AI tasks?
•Typically considered “human”
– Mundane, expert, physical (complex)
Copyright R. Weber
AI tasks
Copyright R. Weber
AI tasks (complex)
• reading &
understanding
• diagnosis
• configuration
• categorization
• classification
• creativity
• discovery
• speech
recognition &
synthesis
• obstacle
avoidance
• NL generation
• NL
understanding
•
•
•
•
•
•
•
•
planning
scheduling
design
prediction
control
monitoring
analysis
vision
Copyright R. Weber
Types of AI tasks
• mundane:
– face recognition
– argumentation
– shopping planning
• expert:
–
–
–
–
diet prescription
medical diagnosis
legal argumentation
legal, military, business planning
• Solution oriented:
– Knowledge discovery
– Text mining
Copyright R. Weber
AI tasks and AI problems
• AI problem is natural language, whereas
related AI tasks are composition, speech,
reading and understanding
• Examples of AI problems can be mechanical
or medical diagnosis and the AI task in both
is diagnosis
• An AI problem is one that requires the
performance of one or many AI tasks to be
solved
Copyright R. Weber
AI applications
Copyright R. Weber
a complete demo
http://www.sls.lcs.mit.edu/sls/whatwedo/applications/jupiter.html
1-888-573-8255
Copyright R. Weber
applications of AI (i)
• knowledge based systems
– financial advise
– medical diagnosis, credit analysis
– case-based reasoning systems for forecasting, case
retrieval, prescription of diet, exercise
– knowledge management systems
– creativity, planning, forecast, recommender,
personalization, argumentation, mediation, tutoring
systems (education),
• military applications
– HICAP
– Immersive Training Environments with virtual reality
Copyright R. Weber
applications of AI (ii)
• machine learning, neural networks
– pattern recognition, e.g., penmanship, voice,
brain activity patterns of motor areas
http://www.txtwriter.com/Onscience/Articles/ratrobot.html
– analyze trends in any domain, knowledge
discovery in databases and text, e.g., data
mining
– classification and clustering and adaptation,
collaborative filtering (movielens)
Copyright R. Weber
applications of AI (iii)
• problem solving by search
–genetic algorithms, games
• robotics, vision
–search for survivors after earthquakes
• recognize survivors
• obstacle avoidance
http://www.nsf.gov/od/lpa/news/03/pr0391_video1.htm
-recognizing criminals in random public locations
–unmanned airplanes, search for land mines
–intelligent user interfaces
Copyright R. Weber
applications of AI (iv)
• natural language
– natural language interfaces
– machine translation
http://babelfish.altavista.com/
– text understanding to analyze patterns &
trends
– summarization, information extraction
• uncertainty
– represent and reason with elements that are
vague and uncertain
– use of fuzzy sets and statistical methods
Dialogue-based NL
• http://www.alicebot.org/
• Winner of the Loebner Prize 2000 & 2001.
JUDGE05[02:19:03] MY NAME IS JUDGE05
PROGRAM[02:19:03] It's good to meet you,
PROGRAM[02:19:03] Judge05.
JUDGE05[02:19:16] You too. What is your name?
JUDGE05[02:19:16]
PROGRAM[02:19:16] Bless you.
PROGRAM[02:19:16] My name is ALICE.
JUDGE05[02:19:37] Good morning Alice. Why have you entered the Loebner Prize?
JUDGE05[02:19:37]
PROGRAM[02:19:37] Hello and how are you this morning?
PROGRAM[02:19:37] The Loebner Prize
PROGRAM[02:19:37] is an annual Turing Test sponsored by New York philanthropist Hugh Loebner. Dr. Wallace and ALICE won it in 2000
Copyright R. Weber
• more chatbots
• http://www.alicebot.org/live.html
• more transcripts at
http://loebner.net/Prizef/2001_Contest/loebner-prize-2001.html
Copyright R. Weber
Necessary grounds for
computer understanding
• Ability to represent knowledge and reason
with it.
• Perceive equivalences and analogies
between two different representations of
the same entity/situation.
• Learning and reorganizing new knowledge.
– From Peter Jackson (1998) Introduction to
Expert systems. Addison-Wesley third edition.
Chapter 2, page 27.