AI, Datamining, Databases Flashcards
AI
Study how to create rational, intelligent agents
-ability of the agent to perform a task in an intelligent manner
Machine learning
- sub-area of AI that aims at giving agents the ability to acquire new knowledge
- the usefulness of a new knowledge is measure with respect of its influence on the overall performance measure
Data mining
-usually performed by a human on specific dataset (that is too big or complex to be analyzed by hand, to specific end
Eg give preliminary insight, discover a pattern, extract information, verify assumptions, or try to predict the future
Studying and learning through
-example problems and solutions
-acquired knowledge
-observations
What is intelligence
- the ability to learn or understand or to deal with new or trying situations
- the skill of reason
- the ability to apply knowledge to manipulate ones environment or to think abstractly as measured by objective criteria
John McCarthy
Father of AI
Who is seen as the first to star working on AI
Alan Turing is seen as the first
-he suggested that AI was best researched by programming computers rather than by building machines
Rational agents
- anything that can perceive it’s environment through sensors and acts upon that environment through effectors
- will always choose the action that maximizes the expected value of the performance measure given the perception sequence to date
Performance measure of agent
It is the criteria which determines how successful an agent is
Perception
It is agents perceptual inputs at a given instance
Perception sequence
It is the history of all that an agent has perceived till date
Behaviour of agent
It is the action that agent performs after any given sequence of precepts
Agent function
It is a map from the precept sequence to an action
Search
Systematic exploration of alternatives for reaching a goal
A problem if defined by four items
- Initial state
- Actions
- Goal test
- Path cost
Solution
Is a sequence of actions leading from the initial state to goal state
Human logic
- humans are information processors
- our strength is the ability to represent and manipulate logical information
Learning
Three different problems being involved in learning
- memory
- averaging
- generalization
Two categories in data mining
Predictive tasks
- classification (walk, drive,fraud)
- regression (education, race , age)
Descriptive tasks
- association discovery (bagel, cream cheese)
- clustering (classical, hip hop)
Predictive data mining
Goal: identify, estimate, or predict a class to previously unseen records as accurately as possible -find a model for the class attribute as a function of the values of the other attributes
Dataset
•Ground truth
-labeled dataset
•training set
- subset of the labeled dataset
- used to learn the model
•testing set
-portion of the dataset not used in building the model used to evaluate the model
Confusion matrix
Predicted
| Pos | Neg
———————
Actual Positive | TP | FN
Negative | FP | TN
Database systems
Provide a way of separating the data user from the data itself, with an interface that allows data operations to be performed easily and efficiently
Database management system
Provides a mechanism for data storage, manipulation and retrieval
Relational database
Classifies data into relations, which are normally called tables. The entries of a table are rows or records
SQL
Structured query language