1 / 9

Lecture 26 Introduction to Machine Learning

This lecture explores the learning problem in machine learning, discussing different approaches such as Bayesian learning, decision tree learning, and neural nets. It also covers the reasons why learning is possible and the importance of bias in learning.

Download Presentation

Lecture 26 Introduction to Machine Learning

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Lecture 26Introduction to Machine Learning CSE 573 Artificial Intelligence I Henry Kautz Fall 2001 CSE 573

  2. Road Map • What is the learning problem? • Why is learning possible? • Approaches to machine learning • Bayesian learning • Decision tree learning • Neural nets • Support vector machines • Nearest neighbor methods • Speed-up learning ... CSE 573

  3. Why Machine Learning • Flood of data WalMart – 25 TerabytesWWW – 1,000 Terabytes • Speed of computation versus slowness of programming highly complex systems (telephone switching systems) = 1 line code @ day @ programmer • Desire for customization a browser that browses by itself? • Sheer ignorance how the heck do you identify gene splice sites? CSE 573

  4. The Learning Problem • Learning = improving with experience at • performing some task • with respect to some performance measure • based on experience chess giving out credit cards CSE 573

  5. Kinds of Learning • Supervised vs Unsupervised • Active vs Passive • Classification vs Action • Empirical vs Analytic • Linear vs Non-linear CSE 573

  6. Terminology – Classification Learning • Instance – described by list of attributes (features) • Target function – some function of the instances we would like to learn • value of chess board • whether or not a credit card holder will default • Concept learning – target is just + or - • Hypothesis space – space of all candidate functions that could be learned – may or may not include the actual target function (if not, is only approximate) • Training set – set of instances labeled with the value of the target function • Test set – labeled data used to measure accuracy of learning CSE 573

  7. Why is Learning Possible? • What can we conclude from: [broccoli, no], [hamburger, yes], [asparagus, no], [cake, yes], [cauliflower, no], [bread, yes] • Experience alone never justifies any conclusion about any unseen instance. • Learning occurs when PREJUDICE meets DATA! CSE 573

  8. Bias • The nice word for prejudice is “bias”. • The world is simple • Occam’s razor “It is needless to do more when less will suffice” – William of Occam, died 1349 of the Black plague • MDL – Minimum description length • Concepts can be approximated by conjunctions of predicates • ... by linear functions • ... by short decision trees • ... by something in the hypothesis space that I choose apriori! CSE 573

  9. Next • Bayesian learning • Decision tree learning • Neural net learning CSE 573

More Related