+ All Categories
Home > Documents > An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and...

An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and...

Date post: 31-Mar-2015
Category:
Upload: darlene-whale
View: 212 times
Download: 0 times
Share this document with a friend
Popular Tags:
20
An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences
Transcript
Page 1: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

An Introduction to Artificial Intelligence

Alexander Yates

Temple UniversityComputer and Information Sciences

Page 2: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Popular Conception of Artificial Intelligence

Page 3: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Artificial Intelligence, more realistically

Page 4: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Autonomous Navigation

• Boston Dynamics’ Big Dog (or fly-dog)– 0:30, 1:25

• Self-driving cars– DARPA’s Grand Challenge

“Ghostrider” (DARPA Grand Challenge entry)• 1:00, 2:20

– Google’s self-driving car:

http://www.youtube.com/watch?v=cdgQpa1pUUE

0:30-1:00

Page 5: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

What is Artificial Intelligence?

It’s the science of making machines behave as if they understand the world the way humans do.

Specifically, AI researchers build machines that can

• know (or believe)• reason (or think or infer or draw conclusions)• learn• communicate (talk and listen)

Page 6: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Expert Systems: Machines than can know and reason

Idea: Make a machine that’s better than a human expert.

Early example:• MYCIN: identifies bacteria causing

severe infections, such as bacteremia and meningitis.– It could propose an acceptable therapy in about 69%

of cases, which was better than the performance of infectious disease experts who were judged using the same criteria.

Page 7: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Expert Systems: Machines than can reason

Second example: ChinookSchaeffer et al.: "Checkers Is Solved“ (2007)

– They proved that the best an opponent can do against their “Chinook” system is to draw.

– It had already been world champion since 1994

Interesting Sidebar:Dr. Marion Tinsley (February 3, 1927–April 3, 1995) is considered the greatest checkers player who ever lived. He was world champion from 1955–1958 and 1975–1991. Tinsley never lost a World Championship match, and lost only nine games (two of them to the Chinook computer program) in his entire 45 year career. He withdrew from championship play during the years 1958–1975, relinquishing the title during that time.

Page 8: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Machines that can learn

One of the more recent advances of AI has been the development of systems that can – Learn from training– Learn from experience– Adapt to changing conditions

25-30 years ago, this was crazy.

Now, there are toolkits: e.g., WEKA and LibSVM

Page 9: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Machines that can learn

Some sample applications:

1) Machines that can determine the amount of air pollution, based on satellite readings

2) Software that can detect network intruders based on network traffic patterns

3) Programs that can diagnose tumors based on MRI scans and CAT scans

All of these are available today.

Page 10: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Conversational Agents: Machines that can hold a conversation

• Chomsky chatbot demo• Video of two chatbots

talking with each other• Speech Recognition

– (e.g., Siri)– You’ve probably interacted with this many

times– Right now, so-so accuracy, but it’s going to

change human-machine interactions

Page 11: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Some problems for AI:

1. How can you make a machine know (or believe)?

2. How can you make a machine think (or reason)?

3. How can you make machines perceive and act upon the real world?

Page 12: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Computational Linguistics

Computational linguistics is the area of AI that deals with language.

A brief tour of computational linguistics today:– Machine translation– Question Answering– Information retrieval– Information Extraction

Page 13: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Machine Translation

Machine translation is the task of translating automatically between languages, like a human interpreter would.

• Google translate demo– English Malay Hungarian English– “I think, Watson, that you have put on seven

and a half pounds since I saw you.”• Very hard problem – it’s considered “AI-

complete”• But a solution is worth $$$$$

Page 14: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Information Retrieval

Information retrieval = Web search

(the Google kind)

The academic study focuses on how to best find the most informative document in a large collection.

1996: 2 Stanford students start working on an information retrieval system called “BackRub”

2004: Their company, Google, has its IPO and makes them billionaires.

Page 15: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Question Answering

Question Answering is the task of finding a specific answer to an arbitrary question in English, using the Web.

Watson

Qualim demo

Page 16: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

16

Information Extraction

Definition:The automatic extraction of structured information

from unstructured documents.

Overall Goals:– Making information more accessible to people– Making information machine-processable

Practical Goal: Build large databases from the information contained in text

TextRunner Demo

Page 17: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

How AI fits into Computer Science

AI

Graphics

Human Computer Interaction

Databases

Data Mining

Systems

Expert Systems and Applications

Theory

Software Engineering

Bug Detection,

Fault Analysis

Approximation Algorithms and

SAT Solvers

Autonomous Video Game Characters

Page 18: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Why study or work on AI?

• It’s exciting – everything is always changing!– True in Computer Science, especially true for

AI– Pressure to innovate: we’re so far from true

AI, we have to constantly re-think how we’re doing things

• It’s useful and it works– AI ideas show up everywhere in everyday life

• It can have a huge impact

Page 19: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Impact of AI

Some of AI’s contributions to the world:– Fraud detection by banks

– Control systems for car brakes

– Automatic Zip Code readers used by the U.S.P.S.

– License plate recognition for EZ-Pass

– Logistic planning used by U.S. Army since the first Iraq War (saving more money than ever was invested into AI research by the government)

– Deep Blue, and other games

– Intelligent user interfaces

– Google & web search

– Mars rovers

– …

Page 20: An Introduction to Artificial Intelligence Alexander Yates Temple University Computer and Information Sciences.

Temple UniversityComputer and Information Sciences

Impact of Computational Linguistics

– Web search

– Speech Recognition

– Speech interfaces used in car GPS systems

– Automatic Arabic translators used by the U.S. Army

– Semantic Web applications (eg, Google News)

– Faster desktop search applications

– Siri

– …

– Many more to come (still a young field)


Recommended