CIS 678 Artificial Intelligence problems deduction, reasoning knowledge representation planning learning natural language processing motion and manipulation.

CIS 678

Artificial Intelligenceproblems

• deduction, reasoning• knowledge representation• planning• learning• natural language processing• motion and manipulation• social intelligence• creativity

CIS 678

Learning

• why is it better than pre-programming a solution?

• where is it better than pre-programming a solution?

• what are its shortcomings?

CIS 678

Machine Learning

• Model a real life process by assuming a distribution and attempt to learn parameters

CIS 678

What do we need?

• Knowledge of statistics and probability• Ability to process data• Ability to apply principles of mathematics

Statistics

Math

Computer Science

ML

CIS 678

What can machine learning do?

• Classification– Predicting tax cheats– Quality control

• Association analysis– Encourage likely purchases– Identify unusual combinations

CIS 678

What can machine learning do?

• Regression– Prediction– Charting relationship

• Clustering (unsupervised)– Grouping similar objects– Describing groups

CIS 678

Classification

• discriminant• prediction• pattern recognition

– optical character recognition– fingerprint, face and speech recognition

• compression• outlier detection

CIS 678

Other concepts

• supervised versus unsupervised• predictive versus descriptive• reinforcement learning (game playing)

CIS 678

Probability

• sample space – the universe of possible outcomes

• events – a single outcome

– example: E = the event that we roll a six

• probability – a number that is associated with the chances of a

particular outcome

– example: P(E) = 1/6