Scribd
Upload
37 views

Notes Unit 1-3 Part-I

The document outlines a syllabus for a Machine Learning course taught by Dr. Rahul Dubey, covering topics such as the introduction to machine learning, model optimization, supervised and unsupervised learning algorithms, and ensemble learners. It includes practical applications using Python and emphasizes the importance of ethical considerations and explainable AI. Additionally, it provides a historical context for machine learning and highlights key milestones in the field.

Uploaded by

Mayank Purohit
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
37 views

Notes Unit 1-3 Part-I

The document outlines a syllabus for a Machine Learning course taught by Dr. Rahul Dubey, covering topics such as the introduction to machine learning, model optimization, supervised and unsupervised learning algorithms, and ensemble learners. It includes practical applications using Python and emphasizes the importance of ethical considerations and explainable AI. Additionally, it provides a historical context for machine learning and highlights key milestones in the field.

Uploaded by

Mayank Purohit
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 20

Machine Learning

2150602

DR. RAHUL DUBEY


ASSISTANT PROFESSOR
DEPARTMENT OF CSE
Syllabus
Unit – I

Introduction to Machine Learning: Learning, Traditional Vs Machine Learning,


Types Of ML, Classification and Regression model, Challenges faced by ML, Steps
of developing an ML model, Bias and Variance, Underfitting and Overfitting,
Regularization, Data visualization, Outlier, Testing and validating, K cross

Dr. Rahul Dubey


validation, Hyperparameter tuning, Model Selection.
Unit – II:
Model optimization and Evaluation: Parametric and non- Parametric model, Learner
performance evaluation, confusion matrix, Recall, accuracy, precision, Model
optimization, Cost/Loss Function, Derivative of cost function and non-derivative
cost function, Gradient descent, Mini-batch Gradient Descent (sckit-learn),
Stochastic Gradient descent(sckit-learn), Momentum(sckit-learn).

2
Unit – III:
Supervised Machine Learning Algorithm with python: Model Complexity vs Dataset
Size,
Supervised Machine Learning Algorithms, k-Nearest Neighbors, Linear Regression, RMSE,
Logistic Regression, Log Loss, Support Vector Machine, Hinge Loss, Kernel Trick,
polynomial
Kernal, Decision Trees, Gini impurity.

Dr. Rahul Dubey


Unit – IV:
Ensemble Learner with python: Ensemble learner, Bagging, Pasting , Voting Classifiers,
Out of-Bag, Evaluation, Random Patches and Random Subspaces , Random Forests , Extra-
Trees, Boosting , AdaBoost, Gradient Boosting, Stacking
Unit –V:
Unsupervised Machine Learning with python: The Curse of Dimensionality, Principal
component analysis, Clustering, K-Means, Limits of K-Means, Clustering, DBSCAN.
ElasticNet, Transformer-based models, Explainable ML, Ethical ML, AutoML, Capstone
Project and Case Studies: Build an end-to-end supervised or unsupervised ML pipeline.

3
Text Books
1) Tom M. Mitchell, “Machine Learning”, McGraw-Hill Education
(INDIAN EDITION), 2013

2) Christopher M. Bishop, "Pattern Recognition and Machine Learning", by


Springer, 2007

3)
Dr. Rahul Dubey
Amanda Casari, Alice Zheng, “Feature Engineering for Machine
Learning”, O’Reilly, 2018.

4) Andreas Muller, "Introduction to Machine Learning with Python: A


Guide for Data Scientists", Shroff/O'Reilly; First edition (2016)

4
Artificial Intelligence, Machine
Learning and Deep Learning & Data
Science

Dr. Rahul Dubey

5
What Is Machine Learning?
 Machine Learning is the science (and art) of
programming computers so they can learn from data.
OR
Dr. Rahul Dubey
 [Machine Learning is the] field of study that gives
computers the ability to learn without being explicitly
programmed.
Arthur Samuel, 1959
What Is Machine Learning?
 More engineering-oriented one:
 A computer program is said to learn from experience E
with respect to some task T and some performance
Dr. Rahul Dubey
measure P, if its performance on T, as measured by P,
improves with experience E.
Tom Mitchell, 1997
Analogy
 spam filter is a Machine Learning program that can learn to
flag spam given examples of spam emails (e.g., flagged by
users) and examples of regular (no spam, also called

Dr. Rahul Dubey


“ham”) emails
 In this case, the task T is to flag spam for new emails, the
experience E is the training data, and the performance
measure P needs to be defined; for
 example, you can use the ratio of correctly classified emails.
This particular performance measure is called accuracy and
it is often used in classification tasks.
Traditional Programming

Data
Computer Output
Program

Dr. Rahul
Machine Learning
Dubey

Data
Computer Program
Output
Traditional Programming

Dr. Rahul Dubey


Machine Learning Approach

Dr. Rahul Dubey


To summarize, Machine Learning is great for:

 Problems for which existing solutions require a lot of


hand tuning or long lists of rules: one Machine
Learning algorithm can often simplify code and
Dr. Rahul Dubey
perform better.
 Complex problems for which there is no good solution
at all using a traditional approach: the best Machine
Learning techniques can find a solution.
To summarize, Machine Learning is great for:
 Fluctuating environments: a Machine Learning system
can adapt to new data.
 Getting insights about complex problems and large
Dr. Rahul Dubey
amounts of data.
ML, DL History
 first machine learning • In 1957, Rosenblatt
system was developed in proposed the
the 1950s perceptron
 In 1952, Arthur Samuel • It was good linear
developedDr. Rahul
a program for Dubey
classifier
playing Checkers it was • But in 1962 minsky
learning Term coined told it can not model
“Machine learning” all problems
• Then no work till 1980’s
ML, DL History
 1986 , decision tree • 1981 ,resurgence of
learning, specifically the neural network by
ID3 algorithm
 In the 90s, machine multilayer perceptron
learning embraced and back propagation
Dr. Rahul
statistics to a large extent
 support vector machines
Dubey
algorithm
were proposed • Back propagation is
 In 1997 Another strong key ingredient of
approach today’s neural
 ensembles or boosting network
Eg. Adaboost
ML, DL History
 kernalized version of • Only Hardware
SVM was proposed acceleration was
near 2000s going on
 Convex optimization
Dr. Rahul Dubey
approach is proposed
“Big Mile Stone”
 New Ensemble
Random forest in 2001

ML, DL History
• new era in neural network
called Deep Learning get
popular in 2005 with
Hinton's lecue and Andrew
research work
• In 1994, the first self driving
Dr. Rahul Dubey
car made a road test
• in 1997, Deep
Blue beat the world
champion Gary Kasparov in
the game of chess
• in 2009 we have
Google building self driving
cars
ML, DL History
• Watson, again from IBM,
won the popular game of
Jeopardy; 2014
• In 2015 CNN surpasses the
Dr. Rahul Dubey
human vision system
• In 2015 machine translation
systems
What Next ?
• How to automate the design of Deep learning
architectures
• How to make the things interpretable

Dr. Rahul Dubey


• Explainable AI or XAI or Interpretable AI
Machine Learning Cycle

Dr. Rahul Dubey

You might also like