Artificial Intelligence

Decision Tree

Decision Tree

can be used to visually and explicitly represent decisions and decision making

Decision Tree

Build a _______ for classifying

Decision Tree

It utilizes supervised learning, batch processing of training

Preference Bias

Define a metric for comparing fs so as to determine whether one is better than another

upside down

A decision tree is drawn ______ with its root at the top

condition or internal node

bold text in black of a decision tree represents a ____

branches or edges

tree splits into _____

decision or leaf

The end of the branch that doesn’t split anymore is the _____

Random

Select any attribute at random

Least-Values

Choose the attribute with the smallest number of possible values

Most-Values

Choose the attribute with the largest number of possible values

Max-Gain

Choose the attribute that has the largest expected information gain

Max-Gain

try to select the attribute that will result in the smallest expected size of the subtrees rooted at its children

H

measures the information content or entropy in bits

Low information content

is desirable in order to make the smallest tree because low information content means that most of examples are classified the SAME, and therefore we would expect that the rest of the tree rooted at this node will be quite small to differentiate between the two classifications.

Conditional entropy

is defined as a conditional probability of a class, Y, given a value, v, for an attribute (i.e., question), X.

question

Pr(Y|X=v) what is X?

label

Pr(Y|X=v) what is Y?

answer to the question

Pr(Y|X=v) what is v?

symmetric

information gain is ________

mutual information

other term for information gain

entropy

measurement of uncertainty

Machine Learning

Is said as a subset of artificial intelligence

data and past experiences

Machine learning is the development of algorithms which allow a computer to learn from the ______ and _______ on their own

Arthur Samuel

Machine Learning was introduced by

1959

what year was Machine Learning introduced?

patterns

Machine learning uses data to detect various ______ in a given dataset.

automatically

It can learn from past data and improve ____________.

data-driven

It is a _________ technology.

data mining

Machine learning is much similar to _______ as it also deals with the huge amount of the data.

increment

Need for Machine Learning Rapid _______ in the production of data

complex

Need for Machine Learning Solving ______ problems, which are difficult for a human

Decision

Need for Machine Learning ______-making in various sector including finance

hidden

Need for Machine Learning Finding ______ patterns and extracting useful information from data

Supervised Learning

Classification of Machine learning Classification/Regression/Estimation

Unsupervised Learning

Classification of Machine learning Clustering/Prediction/Association

Reinforcement Learning

Classification of Machine learning Classification/Control/Decision-Making

Data Exploration

The step where we understand the nature of data that we have to work with. In this, we find Correlations, general trends, and outliers.

Data pre-processing

The step where preprocessing of data for its analysis takes place

Train

_____ model: to improve its performance for better outcome of the problem

Test

_____ model: check for the accuracy of our model by providing a test dataset to it.

Deployment

deploy the model in the real-world system

Herbert Simon

“Learning is any process by which a system improves performance from experience.” Who said dis?

Tom Mitchell

"A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E." Who said dis?

Learning

is essential for unknown environments

system construction

Learning is useful as a _________ method

omniscience

when designer lacks _______

reality

expose the agent to ____ rather than trying to write it down

decision mechanisms

Learning modifies the agent's _______ to improve performance

Machine learning

how to acquire a model on the basis of data / experience

probabilities

example of Learning parameters (plural)

Bayesian network graph

example of Learning structure (singular)

clustering

example of Learning hidden concepts

Supervised Learning

Machine Learning Areas Data and corresponding labels are given

Unsupervised Learning

Machine Learning Areas Only data is given, no labels provided

Semi-Supervised Learning

Machine Learning Areas Some (if not all) labels are present

Reinforcement Learning

Machine Learning Areas An agent interacting with the world makes observations, takes actions, and is rewarded or punished; it should learn to choose actions in such a way as to obtain a lot of reward

past experiences of Data feed in

A machine is said to be learning from ____________ with respect to some class of tasks if its Performance in a given Task improves with the Experience

previous knowledge or past experiences

the machine works in a basic conceptual level of looking at the ________

Data

labeled instances <xi, y>, e.g. emails marked spam/not spam

Features

attribute-value pairs which characterize each x

Experimentation Cycle

-Learn parameters (e.g. model probabilities) on training set -(Tune hyper-parameters on held-out set) -Compute accuracy of test set -Very important: never “peek” at the test set

accuracy

fraction of instances predicted correctly

overfitting

fitting the training data very closely, but not generalizing well

Classification

Learning a discrete function: ________

Regression

Learning a continuous function: _________

discrete

Learning a ______ function: Classification (a SL task where output is having defined labels)

continuous

Learning a ______ function: Regression (a SL task output is having a continuous value)

Data Cleaning

Issues: Data Preparation Preprocess data in order to reduce noise and handle missing values

Relevance Analysis

Issues: Data Preparation Remove the irrelevant or redundant attributes

Data Transformation

Issues: Data Preparation -Generalize data to (higher concepts, discretization) -Normalize attribute values

Model construction

describing a set of predetermined classes

class label

Each tuple/sample is assumed to belong to a predefined class, as determined by the ___________

training set

The set of tuples used for model construction is _______

Model Usage

for classifying future or unknown objects

independent

Test set is _______ of training set, otherwise over-fitting will occur

classify

If the accuracy is acceptable, use the model to ______ data tuples whose class labels are not known

Inductive learning Task

Use particular facts to make more generalized conclusions

predictive

A _____ model based on a branching series of Boolean tests

one-stage

These smaller Boolean tests are less complex than a _____ classifier

measure

We first make a list of attributes that we can _______

discrete

These attributes of the decision tree (for now) must be ______

target attribute

We then choose a _______ that we want to predict

experience table

Then create an ____________ that lists what we have seen in the past

Ross Quinlan

Who developed the ID3 algorithm in 1975?

entropy

ID3 splits attributes based on their ______

entropy

________ is the measure of disinformation

minimized

Entropy is ______ when all values of the target attribute are the same

maximized

Entropy is _______ when there is an equal chance of all values for the target attribute (i.e. the result is random)

lowest

ID3 splits on attributes with the ______ entropy

pruning

There is another technique for reducing the number of attributes used in a tree

prepruning

we decide during the building process when to stop adding attributes

postpruning

waits until the full decision tree has built and then prunes the attributes

expected entropy

ID3 is not optimal because it uses ________ reduction, not actual reduction

errors propagating

Decision trees suffer from a problem of ________ throughout a tree

discretization

We can use a technique known as discretization where We choose cut points for splitting continuous attributes

boundary point

where two adjacent instances in a sorted list have different target value attributes

Lionhead Studios

Black & White was developed by _____ that used ID3

Black & White

Used to predict a player’s reaction to a certain creature’s action