Introduction to
Machine
Learning
What is machine learning
Machine learning is a type of
Artificial Intelligence that
allows machines to learn
from data and improve the
performance without being
explicit programming.
Why use machine
learning?
Machine learning allows the
user to feed a computer
algorithm for process the
amount of data and analyze
and make data-driven
recommendations and decision
based on only the input data.
How it works
ML system use neural networks
and deep learning to collect
data, analyze data, identify
patterns and make predictions.
ML process the data such
images, documents and
numbers.
How it’s difference from AI?
Artificial intelligence is a Machine Learning is a
tool that mimics human subset of AI that focuses on
processing and analytic teaching Machines how to
skills. perform specific task.
AI in every day life : ML in every day life:
Smart products Speech and Image
Map applications recognition
Natural language Traffic alerts
processing(NLP) Chatbot
Machine Learning(ML) Google translation
Robotics, Self-driving cars
Speech recognition and Ads recommendations
translations.
Real world examples
Facialrecognition (verify peoples)
Voice assistants (Google assistant, Alexa,
Iphone’s Siri)
Image recognition (objects, faces,
landmarks)
Email filtering (filtering spam)
Healthcare (mammograms)
Direction apps (Google map)
Recommendations (e-commerce websites)
Types of Machine Learning
There are four types of machine
leaning. They are :
Supervised learning
Semi – supervised learning
Unsupervised learning
Reinforcement learning
Supervised learning
Based on supervision.
It means that the machine using the
labelled data set, and based on the training
the machine predicts the output.
In other word, the machine to be trained
with the input and with corresponding
output, then we can ask the machine to
predict the output using test dataset.
Categories of supervised
learning
Classification (To solve
classification problem like yes or
no, male or female)
Regression (To solve linear
relationship between the input and
output)
Supervised Learning Algorithms :
K-Nearest Neighbors
Linear Regression
Logistic Regression
Support Vector Machines (SVMs)
Decision Trees and Random
Forests
Neural networks
Applications of supervised
learning
Image segmentation
Medical diagnosis
Fraud detection
Spam detection
Speech recognition
Unsupervised learning
As it’s name suggests, there is no need
for supervision.
The main aim of the unsupervised
leaning it to group or categories the
unsorted dataset according to the
similarities, patterns, difference.
That the machines are instructed to find
the hidden pattern from the input
dataset.
Categories of
Unsupervised learning
Clustering (Group the object
into cluster)
Association (To find the
relationship among the
variables within a large
dataset)
Unsupervised Learning Algorithms :
Clustering : K-Means, DBSCAN,
Hierarchical Cluster Analysis (HCA)
Anomaly detection and novelty detection
: One-class SVM, Isolation Forest
Visualization and dimensionality
reduction : Principal Component Analysis
(PCA),Kernel PCA, Locally-Linear
Embedding (LLE) , t-distributed
Stochastic Neighbor Embedding (t-SNE)
Association rule learning : Apriori, Eclat
Applications of
Unsupervised learning
Network analysis
Recommendation systems
Singular value decomposition
Semi - supervised learning
It lies between the supervised and
Unsupervised learning.
It uses the labelled and unlabelled
dataset during the training period.
The main aim is to effectively use all the
available data.
It is highly efficient.
Applications of Semi -
supervised learning
Text document classification
Medical imaging
Natural language processing
Spam email detection
Web content classification
Audio and image analysis
Reinforcement learning
Reinforcement learning works on a
feedback – based process.
The AI agent automatically explore
it’s surrounding by taking action,
learning from experience and
improve it’s performance.
There is no labelled data and the
agent learn from their experiences
Categories of
Reinforcement learning
Positive reinforcement learning
(increasing the tendency by
adding something)
Negative reinforcement learning
(increasing the tendency by
avoiding the negative condition)
Applications of
reinforcement learning
Video games
Resource management
Robotics
Text mining