AI - UT Computer Science

Download Report

Transcript AI - UT Computer Science

Artificial Intelligence
What is AI?
• Turing’s definition
• Are we there yet?
• Eliza (the original)
• Alice
• Joan
Our Working Definition of AI
Artificial intelligence is the study of how to make
computers do things that people are better at or would be
better at if:
• they could extend what they do to a World Wide
Web-sized amount of data, and
• not make mistakes.
A Time Line
View the time line
The Origins of AI Hype
1950 Turing predicted that in about fifty years "an
average interrogator will not have more than a 70
percent chance of making the right identification after five
minutes of questioning".
1957 Newell and Simon predicted that "Within ten
years a computer will be the world's chess champion,
unless the rules bar it from competition."
The Key Components
• Perception
• Knowledge
• Reasoning
• Language
• Moving around
Image Perception
424d961d0300000000003e00000028000000b2030000a5060000010001
0000000000581d0300232e0000232e0000020000000000000000000000
ffffff00e0a288208a38a388a08a00a2880080380288200a38a0082080
380380a00a00a288038a380380000000a00a0080380280a00a00a00800
8a380280a00a00a0000000000000a00a00a0000000000000000000a000
0380380380eb8e00e380e80e38e38abf8e00e38aab8e380380a80a38a3
88abfe3fffffc000e1c71c71c775c71c71c701c71c01c074071c700775
e01c71c0740700700701c71c01c7740700000001e01c01c07407007007
01e01c01c7740700700701e0000000000000700701e000000000000000
0001e00001c074070071c701c700700775c71c7fc701c71c71c7740700
700775c71c71ff7fffffc000e082082082208208208200820800802002
08200220a008208020020020020082080082200200000000a008008020
0200200200a0080082200200200200a0000000000000200200a0000000
000000000000a0000080200200208200020020022002082a0200820820
82200200200220820820aa2aaaabc000e1111011101111101111011110
0100100110101111101011001001001011011114011110010000000150
1001001001001011011010011110010010110110000000000000101101
1000000000000000000110000100100100110000010010101001101500
010110110010010010101001101110151511c000e00000000000000000
0000000000000000000000000000000000000000000000000000000000
0000000000000000000000000000000000000000000000000000000000
Image Perception
Image Perception
http://www.youtube.com/watch?v=t2fjcmPII30&feature=related
http://www.pcworld.com/article/224007/google_wont_release_awesome_facial_reco
gnition_app.html
But We’re Still Ahead
http://www.captcha.net/
But What About Knowledge?
• Why do we need it?
Find me stuff about dogs who save people’s lives.
• How can we represent it and use it?
• How can we acquire it?
But What About Knowledge?
• Why do we need it?
Find me stuff about dogs who save people’s lives.
Two beagles spot a fire.
Their barking alerts
neighbors, who call 911.
• How can we represent it and use it?
• How can we acquire it?
Even Children Know a Lot
A story described in Charniak (1972):
Jane was invited to Jack’s birthday party. She wondered if
he would like a kite. She went into her room and shook her
piggy bank. It made no sound.
Even Children Know a Lot
A story described in Charniak (1972):
Jane was invited to Jack’s birthday party. She wondered if
he would like a kite. She went into her room and shook her
piggy bank. It made no sound.
Replace “kite” by “theorem”, “war”, “water”, …
Replace “piggy bank” by “pillow”, “teddy bear”, “cat”, …
Replace “sound” by “squeak”, “dinner”, “fun”, …
We Divide Things into Concepts
• What’s a party?
• What’s a kite?
• What’s a piggy bank?
What is a Concept?
Let’s start with an easy one: chair
What is a Concept?
Let’s start with an easy one: chair
Merriam Webster:
Chair: a seat …
Seat: a special chair …
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
Chair?
The bottom line?
Can Computers Learn Concepts?
Start with telling the computer that “pie” is a baked good.
Now scan the internet to acquire the concept “baked
goods”:
•
•
•
•
•
Bake the pie for an hour.
Bake cookies at 350 for 14 minutes.
Bake the cake at 350 until a toothpick comes out clean.
Chocolate cake is usually the favorite.
Oatmeal cookies freeze well.
A good notion of
“baked goods”
Can Computers Learn Concepts?
Start with telling the computer that “pie” is a baked good.
Now scan the internet to acquire the concept “baked
goods”:
•
•
•
•
•
•
Bake the pie for an hour.
Bake cookies at 350 for 14 minutes.
Bake the cake at 350 until a toothpick comes out clean.
Chocolate cake is usually the favorite.
Oatmeal cookies freeze well.
I don’t want any Internet cookies.
Can Computers Learn Concepts?
Start with telling the computer that “pie” is a baked good.
Now scan the internet to acquire the concept “baked
goods”:
•
•
•
•
•
•
•
Bake the pie for an hour.
Bake cookies at 350 for 14 minutes.
Bake the cake at 350 until a toothpick comes out clean.
Chocolate cake is usually the favorite.
Oatmeal cookies freeze well.
I don’t want any Internet cookies.
I deleted all my Internet cookies.
Can Computers Learn Concepts?
Start with telling the computer that “pie” is a baked good.
Now scan the internet to acquire the concept “baked
goods”:
•
•
•
•
•
•
•
•
Bake the pie for an hour.
Bake cookies at 350 for 14 minutes.
Bake the cake at 350 until a toothpick comes out clean.
Chocolate cake is usually the favorite.
Oatmeal cookies freeze well.
I don’t want any Internet cookies.
I deleted all my Internet cookies.
I deleted all my files.
Further Complications from How
Language is Used
• After the strike, the president sent them away.
• After the strike, the umpire sent them away.
The word “strike” refers to two different concepts.
When Other Words in Context Aren’t
Enough
• I need a new bonnet.
• The senator moved to table the bill.
Language Processing Today
• Newsblaster
• Watson
Watson
Watch the first day: http://www.youtube.com/watch?v=4PSPvHcLnN0
How does Watson win? http://www.youtube.com/watch?v=d_yXV22O6n4
Machine Translation
The estimated death toll from Japan's disasters climbed past 10,000 Sunday as
authorities raced to combat the threat of multiple nuclear reactor meltdowns and
hundreds of thousands of people struggled to find food and water. The prime
minister said it was the nation's worst crisis since World War II.
El número de muertos por los desastres de Japón subió últimos 10.000 el
domingo como las autoridades se apresuraron a combatir la amenaza de
múltiples fusiones de reactores nucleares y cientos de miles de personas
lucharon para encontrar comida y agua. El primer ministro dijo que era
peor crisis del país desde la Segunda Guerra Mundial.
The death toll from Japan's disaster climbed past 10,000 on Sunday as
authorities scrambled to combat the threat of multiple mergers of nuclear
reactors and hundreds of thousands of people struggled to find food and water.
The prime minister said it was worse crisis in the country since the Second
World War.
Machine Translation
The estimated death toll from Japan's disasters climbed past 10,000 Sunday as
authorities raced to combat the threat of multiple nuclear reactor meltdowns and
hundreds of thousands of people struggled to find food and water. The prime
minister said it was the nation's worst crisis since World War II.
当局は、複数の原 子炉のメルトダウンと食料と水を見つけるのに苦労何十万
もの人々の脅威と戦うためにレースと しての日本の災害からの推定死亡者数
は過去1万日 曜日に登った。首相は、第二次世界大戦以来、国 家の最悪の危
機であると言いました。
Officials estimated the death toll from the disaster of Japan as a race to combat
the threat of hundreds of thousands of people struggling to find food and water
to more than one nuclear reactor meltdown Sunday climbed past 10,000 . The
prime minister since World War II, said the nation's worst crisis.
TranslationParty
A federal appeals court ruled on Tuesday that regulators had limited power
over Web traffic under current law. The decision will allow Internet service
companies to block or slow specific sites and charge video sites like YouTube
to deliver their content faster to users.
連邦控訴裁判所は27日、規制当局は、Webトラフィッ ク過電流法の下で限ら
れた力を持って判決を下した。決定インターネットサービス会社がブロック
する、または特定のサイトやYouTubeのような無料動画 サイトを遅らせる
fasterユーザーにコンテンツを提供するようになります。
A federal appeals court on April 27, regulators, Web has ruled under a limited
power over current traffic laws. The decision to block Internet services
company, or specific sites and free video sites like YouTube to delay the faster
will now provide content to the user.
April 27, regulators, Web, has limited power under the control of traffic control
in the current federal appeals court. Decisions, Internet, and a specific site or
YouTube, the delay can be blocked immediately Bideokontentsusaito as the
current user is free.
http://translationparty.com/
Language Requires Knowledge
Headlines
Beijing Stops Construction for Olympics
Language Requires Knowledge
Headlines
Beijing Stops Construction for Olympics
Hospital Pays $100,000 to Overdose Victim’s Family
Language Requires Knowledge
Headlines
Beijing Stops Construction for Olympics
Hospital Pays $100,000 to Overdose Victim’s Family
TV ads boost eating of obese children.
Expert Systems
Expert Systems
The classic notion: Programs that:
• perform expert tasks, and
• are structured in a particular way:
(as a collection of rules plus an inference engine).
Expert Systems
A broader notion: Programs that:
• perform expert tasks, and
• are structured in a particular way:
(as a collection of rules plus an inference engine).
Expert Systems - Medicine
Example problems:
• arrhythmia recognition from electrocardiograms
• coronary heart disease risk group detection
• monitoring the prescription of restricted use antibiotics
• early melanoma diagnosis
• gene expression data analysis of human lymphoma
• breast cancer diagnosis