Scribd
Upload
87 views23 pages

Statistical Methods in Artificial Intelligence CSE471 - Monsoon 2015

This document provides an overview of a course on statistical methods in artificial intelligence. The course covers topics like linear classification, neural networks, Bayesian classifiers, clustering techniques, and decision trees. Assessment will include assignments, exams, and a project. The goal is for students to understand pattern recognition techniques and apply them to practical applications in areas like computer vision, healthcare, speech processing, and more. Example problems that will be explored include image-based fish classification and challenges like extracting invariant features and handling noise and segmentation.

Uploaded by

srikanth.mujjiga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
87 views23 pages

Statistical Methods in Artificial Intelligence CSE471 - Monsoon 2015

This document provides an overview of a course on statistical methods in artificial intelligence. The course covers topics like linear classification, neural networks, Bayesian classifiers, clustering techniques, and decision trees. Assessment will include assignments, exams, and a project. The goal is for students to understand pattern recognition techniques and apply them to practical applications in areas like computer vision, healthcare, speech processing, and more. Example problems that will be explored include image-based fish classification and challenges like extracting invariant features and handling noise and segmentation.

Uploaded by

srikanth.mujjiga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 23

Statistical Methods in Artificial

Intelligence
CSE471 - Monsoon 2015

Avinash Sharma
CVIT, IIIT Hyderabad

Course Content

Introduction
Linear Classification
Neural Networks
Probability Densities
Bayesian Classifiers
Dimensionality Reduction
Support Vector Machines
Kernel Methods
Clustering Techniques
Decision Tree/Graphical Models

Reference Material
Books
Pattern Classification by Duda, Hart & Stork
The Elements of Statistical Learning by Hastie, Tibshirani and
Friedman
Machine Learning : A probabilistic Perspective by Kevin P. Murphy

Pre-requisite
Basics of Linear Algebra, Probability Theory and Statistics.
Programming in Matlab and C/C++.

Course Website

http://courses.iiit.ac.in

Online Courses/Tutorials and Research Papers

Assessment & Attendance


Policy
Assessment
27% Project
18% Assignments (Best of 6 out of 7)
30% Two Mid-semester Exams (15% each)
25% Final Exam
Zero Marks in assignment to both parties in case
a copy is detected
Cooperation with TAs is expected

Zero Tolerance for institute attendance policy

Expected Outcome
This course would enable students to
understand pattern recognition techniques in
detail.
We will ensure that both theoretical as well as
practical aspects are learnt simultaneously.
The project deliverables are expected to be
working systems attached to some practical
application.

Introduction
Statistical Methods
Frequentist Approach
Probabilistic Approach

Artificial Intelligence
Pattern Recognition (PR)

Our focus is PR with SM in AI


Key Problems in PR
Clustering
(Seeking natural division of data)
Classification (Assignment to category labels)

Classification
Preprocessing: It involves cleaning the input
data for noise and clutter.
Feature Extraction:
Each property of a data
instance/observation/sample is termed a feature,
also known in statistics as an explanatory variable.
A feature vector is typically a multi-dimensional
vector that stores individual and measurable
properties of each of data instance.
Features can be Binary, Categorical or Real-valued .

Classification
Identification of a specific category label from a
set of categories a new observation in the test
set of data belongs to.
This is achieved using the training set of data
containing observations whose category
membership is already known.
Both training and test set of data samples are
assumed to be drawn form the same
distribution.

Applications

Vision/Robotics
Healthcare
Life Sciences
Text Analytics
Speech Processing
Business
Surveillance

Applications

Vision/Robotics
Healthcare
Life Sciences
Text Analytics
Speech Processing
Business
Surveillance

Diseased

Healthy

Applications

Vision/Robotics
Healthcare
Life Sciences
Text Analytics
Speech Processing
Business
Surveillance

Toy Examples Walkthrough


Image Based Fish Classification
v/s Sea Bass)

(Salmon

Toy Examples Walkthrough


What could be the best features
representing these two fish?
Length
Width
Appearance
Lightness (reflective or shiny skin)
Texture

Fin Structure

Toy Examples Walkthrough

ghtness Histogram Feature

Length Histogram Featu

Toy Examples Walkthrough

Toy Examples Walkthrough

Toy Examples Walkthrough

PR System Flow

Challenges
Segmentation
Recovering the Signal from noise/clutter

Feature Extraction
Invariance to Translation, Rotation and
Scaling
Occlusion
Rate
Deformation

Challenges
Classification
Noise Any property of the sensed pattern
which is not due to true underlying model
but instead to randomness in the world or
the sensor.

Post Processing
Error Rate
Risk
Context
Multiple Classifiers

PR System Design Cycle

Types of Learning
Supervised Learning
Unsupervised Learning
Reinforcement Learning

Next Lecture
Linear Discriminant Functions
(Chapter 5)

Thanks

You might also like