how do you get here? https:// youtube/watch?v=t9fxp3hk6di

31
• How do you get here? https:// www.youtube.com/watch?v=t9Fxp3HK6DI

Upload: conan-hooper

Post on 03-Jan-2016

37 views

Category:

Documents


0 download

DESCRIPTION

How do you get here? https:// www.youtube.com/watch?v=t9Fxp3HK6DI. Pattern Recognition & Machine Learning. Patterns. Humans are excellent at recognizing patterns. Patterns. Even if we can't explain how we do it…. Nearest Neighbor. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

• How do you get here?https://www.youtube.com/watch?v=t9Fxp3HK6DI

Page 2: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Pattern Recognition& Machine Learning

Page 3: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Patterns

• Humans are excellent at recognizing patterns

Page 4: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Patterns

• Even if we can't explain how we do it…

Page 5: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Trick 1: Nearest Neighbor

• Task : predict what houses are most likely to donate to an election

Page 6: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Nearest Neighbor

• Task : predict what houses are most likely to donate to an election

• Know some voter• registrations

Page 7: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Nearest Neighbor

• Task : predict what houses are most likely to donate to an election

• What should wepredict for the ?marks

Page 8: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Nearest Neighbor

• Task : predict what houses are most likely to donate to an election

• Should we considermore than oneneighbor?

Page 9: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Simulator:• http://www.cs.cmu.edu/~zhuxj/courseproject/knndemo/KNN.html

Page 10: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Simple Nearest Neighbor

• Nearest Neighbor Applied

Pattern Nearest Neighbor Nearest 3 Neighbors

Page 11: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Other Nearest Neighbor

• Nearness as pixel difference:

Page 12: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Trick 2: Decision Trees

• Sequnce of choices to make a decision

Do I need an umbrella?

Page 13: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Spam Filter

• Is a web page "spam"?

Page 14: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Spam Filter

• Is a web page "spam"?

Page 15: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Spam Filter

• Is a web page "spam"?

How do we decide the questions???

Page 16: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Machine Learning

• Machine Learning : Build a general algorithm to LEARN specific patterns

Page 17: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Learning a Decision Tree

• http://aispace.org/dTree/

Page 18: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Human Involvement

• Still need to determine possible questions, things to look at

Page 19: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Human Involvement

• Still need to determine possible questions, things to look at– What should we look at for these???

Page 20: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Trick 3: Neural Networks

• Biologically inspired computation

Page 21: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Neural Networks

• Biologically inspired computation

Page 22: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Neural Networks

• A simple "take umbrella" network:

Page 23: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Neural Networks

Page 24: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Sunglasses Network

• Image recognition network:

Page 25: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Sunglasses Network

• Image recognition network:

Page 26: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Enhanced Neurons

• Signals can be any value 0-1

Page 27: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Enhanced Neurons

• Signals can be any value 0-1• Inputs can be weighted

Page 28: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Enhanced Neurons

• Signals can be any value 0-1• Inputs can be weighted• Threshold function is not all or nothing– Produces values 0-1

Page 29: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Learning

• http://aispace.org/neural/

Page 30: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Result

• One neuron's weights

Page 31: How  do you get  here? https:// youtube/watch?v=t9Fxp3HK6DI

Making it all worth it

• http://www.cs.cmu.edu/~tom7/mario/