You are on page 1of 13

Introduction to Machine Learning

Dr. Shekha Rai

Department of Electrical Engineering


National Institute of Technology Rourkela
India

1 / 13
Syllabus

Introduction
Different types of learning
Linear regression
Decision trees
KNN
PCA
Bayesian learning
Logistic Regression
SVM
ANN
Introduction to ensembles
Bagging and Boosting
K-means clustering
2 / 13
Books recommended

Tom M Mitchell, Machine Learning, PHI , 2015

Ethem Alpaydin, Introduction to Machine Learning, The MIT


Press , 3rd Edition 2015

Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani,


Introduction to Statistical Learning, Springer, 2013

Richard Duda, Peter Hart, David Stork, Pattern Classification,


John Willey

Saikat Dutt, Subramanian Chandramouli, Amit Kumar Das,


Machine learning, Pearson

3 / 13
History

1950- Alan Turing proposes "learning machine."

1952- Arthur Samuel first developed machine learning program


that could play checkers.

1957- Frank Rosenblatt designed the first neural network


program simulating the human brain.

1969- Nearest neighbour algorithm.

1982- Recurrent neural network.

1995- Suport vector machine and random forest algorithm.

2016- Google’s alpha program has beaten professional human


player using ML techniques.

4 / 13
Defination

Enables computers to think and learn on their own.

The machine gets trained on a training dataset, large enough to


create a model, which helps it to take decisions based on its
learning.

5 / 13
Defination

Tom M.Mitchell has defined ML as


A computer program is said to learn from experience ’E’ with
respect to some class of task ’T’ and performance measure ’P’, if
its performance at task in ’T’, as measured by ’P’, improves with
experience ’E’.

6 / 13
Example

Image classification problem:

Past data with images having labels.

Assigning labels to new unlabeled images.

% of images correctly classified.

7 / 13
Example

Image classification problem:

E→ Past data with images having labels.

T→Assigning labels to new unlabeled images.

P→ % of images correctly classified.

8 / 13
Process of ML

Figure: Process of ML

9 / 13
Process of ML

1 Data i/p: Past data or information is utilized as a basis for future


decision making.
2 Abstraction: The input data is represented in a broader way
through an underlying algorithm.
This model may be of the form
Computational block like if/else rules
Mathematical equations
Specific data structures like trees/graphs
Logical grouping of similar observations.

Choice of model to solve a specific learning problem is a human


task which is taken on the basis of
Type of problem to be solved- forecasting, analysis of trend,
understanding different groups of objects.
Nature of i/p data- data types.
Domain of the problem- business domain etc.

10 / 13
Process of ML

(2) Abstraction:
Once the model is chosen, the next task is to fit the model based
on the input data- training
Input data based on which the model is finalizes- training data.
(3) Generalization:
Tune up the abstracted knowlege to a form which can be used for
future decisions.
Problems may be encountered when the model is applied to take
decisions on a set of unknown data (test data) .
The test data may possess certain characteristics unknown to the
training data.

11 / 13
Applications of ML

Robot control

Design autonomous robot that learn to navigate from their own


experience

Healthcare

Diagnose a disease

i/p → Symptom: lab measurements, test results

o/p → One of the set of possible diseases

Financial

Predict if a stock will rise or decline

12 / 13
Thank You

13 / 13

You might also like