Bayesian Networks
1
Bayesian Networks
HasAnthrax
HasCough
HasFever
HasDifficultyBreathing
HasWideMediastinum
In the opinion of many AI researchers, Bayesian
networks are the most significant contribution in
AI in the last 10 years
They are used in many
Regression & Regularization
Dr Sattar Hashemi
1/21
Regression
So far, weve been interested in learning P(YX) where Y has
discrete values (called classification)
What if Y is continuous? (called regression)
predict weight from gender, height, age,
pred
Logistic Regression vs. Nave Bayes
Dr Sattar Hashemi
This slides is based on:
Ng, Andrew Y., and Michael I. Jordan. "On discriminative vs. generative classifiers:
A comparison of logistic regression and naive Bayes." Advances in neural
information process
Probabilistic
Temporal Models
Ali Mashhoori
1
Examples
Given an audio waveform, would like to robustly
extract & recognize any spoken words.
2
Examples
Estimate motion of targets in 3D world from indirect,
potentially noisy measurements
Radarbased tracki
Machine Learning
Decision Trees
1
Decision tree
Decision tree representation
Each internal node tests an attribute
Each branch corresponds to attribute value
Each leaf node assigns a classification
Represent AB, A XOR B
2
How
would you represent Boolean
Data Stream
Definition
Ordered sequence of data
Huge volumes , possibly
infinite
Fast changing and
requires fast response
Single linear scan
algorithm: can only have
one look
18 April 2010
2
Data Stream
Challenges
Data
Volume
Concept
Change
18 April
Data Stream Classification
Data Stream
Definition
Ordered sequence of
data
Huge volumes ,
possibly infinite
Fast changing and
requires fast response
Single linear scan
algorithm: can only
have one look
5/26/16
2
Data Stream
Challenges
Data
Volume
Dat
1
2
3
1) Show, using a proof or an example, that if Pr(AB,C) = Pr(AC), then Pr(A,BC) =
Pr(AC) Pr(BC).
2) a) Consider a naive Bayes classifier trained on the dataset from Table 1. How would
that predict Edible given the input Colour = B, Odour = 2? Sh
Data Mining:
Concepts and
Techniques
Chapter 5
Jiawei Han and Micheline Kamber
Department of Computer Science
University of Illinois at UrbanaChampaign
www.cs.uiuc.edu/~hanj
2006 Jiawei Han and Micheline Kamber. All rights reserved.
May 26, 2016
Data M
Data Mining:
Concepts and
Techniques
Chapter 5
Jiawei Han and Micheline Kamber
Department of Computer Science
University of Illinois at UrbanaChampaign
www.cs.uiuc.edu/~hanj
2006 Jiawei Han and Micheline Kamber. All rights reserved.
May 26, 2016
Data M
BAGGING & BOOSTING
1
By: Ali Mashhoori
DEFINITION OF THE PROBLEM
We are given a training set , consisting of
pairs.
is a vector of features.
is either:
A class label: CLASSIFICATION
A real number: REGRESSION
We want to predict for an unseen data point
2
H
Graph Based Data Mining:
Mining Frequent
Substructures
from Graph Data
T. Washio
I.S.I.R., Osaka University
March, 8th, 2004
1
Background
Graph structure also frequently appears in
realworld, e.g., web links and chemical
compounds.
CASE and MultiCASE sys
Graph Based Data Mining:
Mining Frequent
Substructures
from Graph Data
T. Washio
I.S.I.R., Osaka University
March, 8th, 2004
1
Background
Graph structure also frequently appears in
realworld, e.g., web links and chemical
compounds.
CASE and MultiCASE sys
Introduction
Dr Sattar Hashemi
George E. P. Box
1/20
What is a Machine Learning?

Is Machine Learning just Statistics ?
Is Machine Learning just Optimization ?

In 1970, this question was emerged: why the new field is required?
There is nothing being do