Artificial Intelligence in Action

In recent years, AI has made commendable progress. From the first independent
shopping cart that delivers your food products to AI-generated music, AI is
everywhere.

Globally, many organizations have started innovating a variety of tools and

products that enhance business outcomes.

TCS ignio™ is one such innovative cognitive automation solution.

Digitate, a software venture of Tata Consultancy Services, won the Best Overall AI
Company award from AI Breakthrough for 2019.
Evolutionary Levels of AI
Evolutionary Levels of AI

The following levels mark the rise of AI.

Narrow Intelligence
It applies to specific tasks, for example, Alexa, Sophia and Alpha Go.
General Intelligence
Machines are capable of imitating human intelligence.
Super Intelligence
At this level, machines can surpass humans.

If you are worried that there is a human-like intelligence lurking in every

cupboard, breathe easy. All those AI applications are too busy solving real business
problems.
Having had a glimpse of the evolution, you now know that AI has been here for more
than half-a-century, so aren't you curious to find out why it is gaining importance
now?

More Data

With social media coming into existence and with technology gaining momentum in all
fields, data is getting generated at an n-measurable pace triggering the need for
complex processing.

Accurate Algorithms

With data getting collected, now arises the need for algorithms that will perform
accurate and quick computations.

More Computation Power

Much breakthrough has been made in the space of algorithms with the arrival of
advanced machine and deep learning technologies. These algorithms mandate high
computational power.

7 of 15
Research Facts

Four key areas stand out based on a study done to understand the impact of AI in
organizations. The study considered the views of 835 business leaders from companies
across the globe.

Investments in AI and business impact show correlation.

Companies have started to see AI as essential.

The impact is spreading beyond IT departments to other roles such as admin, back
office, and sales and marketing.

Moreover, though the overall view of AI’s impact is positive, there are
important considerations about the effects it may introduce.
Symphony of AI
Symphony of AI

Each instrument must be played with great skill before you can strike that last
magnificent chord.

If you wish to play the AI orchestra, you will need to know in detail about its
primary instruments, which are listed below.

Data Science
Machine Learning
Deep Learning
Natural Language Processing
Artificial General Intelligence

Play on! Your first chord has just begun.

Defining AI

Non-biological entities exhibiting complex, goal-oriented behavior is termed as

Artificial Intelligence.
Many popular products that we use involve the magic of AI. Some of the AI-based
products are listed below.

Google Predictive Search Engine.

Google AI Eye Doctor
Spam Filtering in Gmail
Google Virtual Assistant
Siri
Alexa
Self Driving Cars
Netflix recommendations
Programming Languages of AI

To play the AI chords, you need a good understanding of one of the below notes.
 Python
R
Java
Lisp
C++
Matlab
Julia
JavaScript
100% of the energy companies use AI, which is the only industry to achieve this
feat.
Data on the Rise

There is a plethora of information flowing in from digital devices you use. Every
like or comment you make on social media, every image when clicked, every file when
saved, and every Google search you do, produce data.

Are you wondering how data is going to help? You may have the following questions in
your mind.

What kind of knowledge?

How is insight retrieved?
Which actionable decision is made?

All your queries will be answered by Data Science.

Defining Data Science

Data Science refers to the activity of analyzing a large amount of data to extract
knowledge and insight leading to actionable decisions.

The main components in Data Science are, namely:

Acquiring
Cleaning
Describing
Exploratory Analysis
Making predictions
Suggesting recommendations

Explaining the Components

Acquiring Data

Acquire the raw data from a multitude of data sources, such as RDBMS and web
pages.

Cleaning Data

Convert the raw data into a machine-readable format (enrich, detect/remove

outliers, and apply business rules).

Describing Data

Summarize the data to get a holistic picture.

Exploratory Analysis

Explore and determine the patterns from the data. Visualize the data to unearth
insights.

Making Predictions

Generalize the patterns in data to build models.

Make predictions on unknown data.

Suggesting Recommendations

Now you are good to suggest recommendations.

Apophenia is the tendency to see patterns in random data.Data Science in Real Life

It is time for you to know the power of Data Science.

IBM Watson explained the idea of Data Science in IBM DeepQA research by winning
against the Jeopardy champions, Ken Jennings and Brad Rutter.

Amazon's recommendation engines are suggesting the items for you to buy.

The list goes on with the recommendations from Netflix and Spotify.

Self-driving cars are enabled using algorithms that can detect objects such as
traffic lights, other cars on the road, and pedestrians.

By 2050, it is estimated that there would be 50 billion devices connected to the

internet, creating a hyperconnected age.
Data science must scale across all the layers of the organization. To ensure this ,
the three important elements to consider are:

Data
Architecture
Tools

Data

A proper strategy must be put in place to ensure that data scientists have easy
access to the sources of data.
Data governance must be dealt cautiously.
 Data governance is the overall management of the availability, usability,
integrity, and security of data used in an enterprise.
Architecture

Traditional Monolithic Architecture

Features are bundled in a single deployment location.

Service-Oriented Architecture

Functionalities are drilled down into services that will be deployed independently
to ensure high efficiency and scalability.
Tools

You need the right tools to help you perform data science operations with ease.

Python
R
RapidMiner
DataRobot
Apache Hadoop
Trifacta
Alteryx
KNIME
MATLAB

Industry Standard

You have covered the components of Data Science in the previous cards.

To implement the same in an industry context, you need knowledge on few open
standards such as CRISP-DM (Cross-industry standard process for data mining).

It has six phases, namely:

Business understanding
Data understanding
Data preparation
Modeling
Evaluation
Deployment

Polls conducted in 2002, 2004, 2007, and 2014 shows that the CRISP-DM methodology is
the leading methodology used by data miners.

Prelude

Machine learning is the next instrument in the AI symphony. So what exactly is

Machine Learning?
 Machine Learning is a subset of AI that provides machines the capability to
learn automatically and improve from experience without being explicitly programmed.

In simple words, it helps machines to solve a given problem by gaining the ability
to think.
Knowing the ML Terms

Before diving deep into ML, you have to be aware of the following terms.

Algorithm

Algorithm is a set of rules and statistical techniques used to learn and derive
insights from data patterns. e.g., Decision tree, Linear Regression, and Random
Forest.

ML Model

ML model is a mathematical model trained by an algorithm to predict the patterns in

the data.

Predictor Variable

Predictor variable is a variable used to predict another variable/output.

Response Variable

Response Variable is the target variable or the output variable that needs to be
predicted.

Training Data

A model is built using training data.

Testing Data

Model is evaluated using testing data.

Predictor vs Response Variable

In the scenario where the height of the individual is predicted based on age, the
predictor variable will be Age, and the response variable will be Height.
Defining Your ML Problem

If I had an hour to solve a problem, I'd spend 55 minutes thinking about the problem
and 5 minutes thinking about solutions. - Albert Einstein

The following questions will aid you in defining a problem:

What kind of problem are you facing?

How much or how many? (Regression)
 Which category? (Classification)
Which group? (Clustering)
Is this an outlier? (Anomaly detection)
Which option to opt? (Recommendation)

Once you have defined your problem, you need to strike the data science chord to get
the data prepped up for further processing.
Exploring ML

Let's now dive into the process of Machine Learning!

The ML steps involved in predicting an outcome for a given problem are listed below.

Build the Model

Model Evaluation
Prediction

Exploring ML
Exploring ML

Build the Model

This step starts by splitting the data, which is ideally split into

Training data, and

Testing data

Data Splitting follows the rule-of-thumb 80/20 rule. 80% of data contributes to
Training set while the remaining 20% is assigned to the Testing set.

Training data is used to train (build) the model using algorithms such as Linear
regression and Decision Tree.

If you feed more data in the training phase, the model will perform better in the
testing phase.
Exploring ML

Model Evaluation

It's time to put the model you just built to test.

In this step, using the testing data, you will check the predictions of your model,
based on which you can tune the accuracy and efficiency of the model.

Parameter Tuning is a popular technique that helps in improving the efficiency

of the model.

Prediction

In this stage, the model is all set to give predictions for the input you provide.
A Sample Scenario

Using examples to back up statements of fact can add value to your concept.

Predict the occurrence of rain in your area with the ML process by studying the
weather condition.

Define the Problem

Possibility of rain (yes or no - Response Variable)

Temperature, Humidity condition (Predictor Variable:)

Data Gathering and Cleaning

Data Source: Meteorological Research Center

Exploratory Data Analysis

Insight 1: High probability of rain, if the temperature has fallen low.

Insight 2: When it rains, the humidity is at 100%.

Build the Model

Use suitable ML algorithms to train the model.

Evaluate the Model and then you are good to predict.

Types of ML

Machine Learning can be broadly categorized into three types, as listed below.

Supervised
Unsupervised
Reinforcement Learning

Imagine unsupervised learning as a type of mathematical version of making “birds of

the same feather flock together.”
Supervised vs Unsupervised Learning
Supervised vs Unsupervised Learning

Supervised Learning - A technique that uses labeled data for training the model.

Unsupervised Learning - A technique that uses unlabeled data for training the
model.

The GIF shown here portrays this difference.

In supervised learning, upon giving the labeled pictures as input, the model is
trained to produce two outputs, namely Popeye and Olive.
In unsupervised learning, when unlabeled pictures are given in, they try to extract
the prominent features of the images and group them accordingly.
Reinforcement Learning

Imagine yourself dropped off at an island all alone!

What would you do?

Panic? Yes, of course, I bet you will be.

As time passes, you will learn to live in the environment by exploring the island,
climatic conditions, food available, threats, and safety measures.

As a result of the learning, you will get adapted to the environment by finding out
which fruits to eat and what not to do.

This is an example of Reinforcement Learning, which is defined as,

A part of Machine Learning where an agent is put in an environment where he

learns to behave by performing actions that will either be rewarded or punished.

AlphaGo is Google's AI Reinforcement Learning example.

Deep Learning

Deep learning is a branch of machine learning where algorithms are inspired from
the way neurons work in a human brain.

Artificial neurons form the crux here, which enables the machines to think naturally
like humans.

This is the main building block for technologies such as driverless cars.
Natural Language Processing

Now, you are an inch closer to play the AI music.

So, what is NLP?

NLP - A machine’s ability to understand the human language by breaking it down,

and then comprehend its meaning to determine the appropriate action, and respond in
a language the user will understand.

68% of companies use AI in IT, of which, 66% use it to detect and deter security
intrusions.
Explore More!

Here is the list of courses, you can explore to gain in-depth knowledge in the AI
space.

Machine Learning Axioms

 Machine Learning - Exploring the Model
Clustering - Data Ensemble
Regression Analysis
Association Rule Mining
Neural Networks and Deep Learning
Build Effective Deep Learning Models
Convolutional Neural Networks
Machines in Life

Machines have integrated into our daily life and are now playing a prominent role in
defining our culture.

From the steam engine, electric bulb, computer and to the robots, machines have been
at the heart of the industrial revolution from the late 18th century.

TCS' Machine First™ Delivery approach helps to optimize human-machine collaboration.

Machine First™ Approach

Customer experience drives the digital world's business excellence. The omnipresence
of digital technologies provides businesses and company leaders a myriad of options.

Smart machines will help you to eliminate mundane steps by providing countless
choices. Using these choices, you can make instant decisions that will enhance the
customer experience.

Hence, companies are moving in a direction where machines will be the first to carry
out a variety of operations.
What is MFDM™?

TCS defines Machine First™ approach as,

The Machine First™ approach allows technology the first right of refusal to
sense, understand, decide, and respond in a robust networked environment equipped
with analytics and AI, with the learning platform enabling superior quality
information across the enterprise in real-time, all the time.

The MFDM™ Framework helps organizations to deploy the defined approach seamlessly.
MFDM™ Framework Constitutes

Enterprise Intelligence Platform: Inputs from various sources are converted into a
format that machines can read to make decisions.

Enterprise Response Engine: It orchestrates the decided course of action.

Collaboration Platform: It helps the machines to learn.

MFDM™ describes a maturity model, which helps organizations to map incremental
phases of maturity in the transformation journey to different levels, namely:

Directed
Assisted
 Collaborative
Autonomous

The different phases of execution model that ensure successful delivery using the
Machine First™ approach are listed below.

Discover
Adapt
Transform

Discover

The current state of the enterprise IT landscape is studied in detail and the target
state, the machine first roadmap, the solution design for Enterprise Intelligence
Platform and Enterprise Response Engine are defined.
Adapt

The Enterprise Intelligence Platform and the Enterprise Response Engine are
established.
Transform

A multitude of automation capabilities are identified, continuously developed, and

deployed using cognitive capabilities.
Benefit
1

Drives superior customer experience

Transforms business operations
Customized services for stakeholders
Free human resources for high-value work
Success Speaks
Success Speaks

MFDM™ in
TCS

Energy Management System for TCS facilities

Conversational Intelligence applied to HR

Manufacturing

Recall Management
Smart Inventory

Utilities

Maintenance Predictions
Customer Consumption Profile

Insurance
 Straight through claim processing
Fraud Detection and Prevention

Success Speaks

MFDM™ in
TTH

Intelligent Freight
Dynamic Pricing

Retail

Demand Forecasting
Customized Promotions

Banking

Algorithmic Trading
AI Assisted Investments

Communications

Network Optimization
Real Time Customer Data Analysis

Healthcare

Virtual Consultation
Patient Data Management

TCS AI Products

Prominent AI products and platforms of TCS:

ignio™

Optumera™

Advanced Drug Development Platforms

Challenge

In today's business IT environment, services depend mainly on human knowledge,

intuition, and experience. The abundance of data and information in an enterprise
makes it complicated and time-consuming to apply intelligence and decision making.
ignio™
ignio™
TCS worked on developing a robust intelligent automation tool that is capable of
mimicking the human thought process.

ignio™

ignio™ is a cognitive automation product for enterprise IT operations.

It was launched in June 2015.
It dynamically combines Machine Learning, Artificial Intelligence, and
Automation.

ignio™ as defined by TCS

ignio™ is an open, extensible platform that is specifically designed to optimize

your time while helping you make smart business decisions. The system is dynamic,
and it assimilates additional capabilities as your IT ecosystem continues to grow.

Key Features of ignio™

Proactive

ignio™ warns the user for any potential incident and thus empowers the user to
take appropriate action.

Agile

ignio™ helps in automating repetitive tasks.

It enhances productivity and end-user experience.
It measures the system's health periodically to initiate corrective actions.
It enables quick decisions and provides recommendations for better performance.

Reliable

ignio™ can identify probable causes and eventually recover from outages or
incidents within seconds.

Four Key Products

Artificial Intelligence Operations (AI Ops)

Intelligent automation (IA) on datacenter activities such as infrastructure and

application operations.

AI Digital Workspace

Intelligent Automation on end-user devices such as desktops and mobile devices.

AI Workload Management

It manages batch operations.

AI ERP Ops

AI ERP Ops caters to key business processes that are implemented through systems
like SAP.

Numbers to Wonder

Wondering what is possible with ignio™? Here are some numbers for you:

ignio™ improves business availability through 80-90% reduction in operational

risks and outages.
It provides better end-user experience through 90% reduction in time to perform
activities.
It delivers a high ROI through 50-60% of reduction in effort and cost.

Optumera™
Optumera™

TCS defines Optumera™ as

Tata Consultancy Services' (TCS') Optumera suite is a next-generation

merchandising solution that helps retailers reinvent their businesses by embracing
the evolving consumer, technology and market trends, and implementing cutting-edge
merchandising strategies – placing the shopper in the driver's seat.
Preamble

The concepts of artificial intelligence and machine learning always evoke the
ancient Greek myth of Pandora’s box. In the fairytale version of the story, Pandora
is portrayed as a curious woman who opened a sealed urn and inadvertently released
eternal misery on humankind.

In the original telling, Pandora was not an innocent girl who succumbed to the
temptation to open a forbidden jar. Instead, as the poet Hesiod tells us, Pandora
was made, not born.

Like the genie that escaped the lamp, the horse that fled the barn, the myth has
become a cliche. Now, let us explore the Machine Learning to get more fascinated!
Data Everywhere!

We are drowning in information and starving for knowledge.

Google

24 petabytes of data are processed per day.

Facebook

10 million photos are uploaded every hour.

Youtube

1 hour of video is uploaded every second.

Twitter

400 million tweets are posted per day.

With data increasing every day, we can believe that smart data analysis will become
more prevalent as a fundamental ingredient for technological progress.
Why Machine Learning?

We interact with Machine Learning models every single day without our knowledge.

Every time we Google something, or listen to a song or even take a photo, Machine
Learning is becoming a backbone process behind it by invariably learning and
improving from every interaction.

Machines can drive your car for you, detect eye diseases, unlock your phone with
face recognition, and the list never ends.

Let us get started with Machine Learning!

What is Machine Learning?
What is Machine Learning?
Definition

Machine Learning is the field of study that gives computers the ability to learn
without being explicitly programmed.

Machine learning is a tool for turning information into knowledge.

We are DATAFIED! Wherever we go, we leave a data trail. Data becomes fruitless
unless we discover the hidden patterns. Wondering how? Yes! Machine Learning is a
magic wand that turns information to knowledge, which will do wonders for humankind.

Deep dive into the concepts to know more.

Traditional Learning

Blends human-created rules with data to create answers to a problem.

Machine Learning

Uses data and answers to uncover the rules that build a problem.

What Machine Learning does?

Do you want to predict a category?

Machine Learning has Classification

Example

Predict if the stock price will increase or decrease.

Do you want to predict a quantity?

Machine Learning has Regression

Example

Predict the age of a person based on their height, weight, and health factors.
What Machine Learning does?

Do you want to detect an anomaly?

Machine Learning has Anomaly Detection

Example

Money withdrawal anomalies can be discovered.

Do you want to discover structure in unexplored data?

Machine Learning has Clustering

Example

Finding a group of customers with similar behavior based on their buying data
history.
Prelude

While a great deal of engrossment has been towards model building, model tuning, and
model evaluation, many individuals still find themselves asking basic inquisitive
questions like

What is the life cycle of Machine Learning?

This section of the course will aid in answering this question. Keep reading to know
more!
Big Picture

The big picture of Machine Learning process lies in the following 9 steps namely

Defining Project Objectives

Gathering Data
Exploratory Data Analysis(EDA) and Data Cleaning
Choosing a Model
Training
Evaluation
Hyperparameter Tuning
Interpret and Communicate
Deployment and Documentation

Defining Project Objectives

The first step of the life cycle is to recognize the opportunity for tangible
improvement of activities, to enhance customer satisfaction, or to create value
otherwise.
It is critical that you understand the problem you are trying to solve. In this
stage, you should also be identifying the central objectives of your project by
identifying the variables that need to be predicted.

Gathering Data

Considered to be the primary step of Machine Learning process.

The quality and quantity of data you gather in this step will determine how
efficient your model will be.

Some important things to remember while gathering data:

Data can be collected from anywhere in any format.

More training examples will aid the model to be more efficient.
Make sure the number of samples for every class or topic is not overly
imbalanced.
 Make sure that your samples adequately cover the space of possible inputs, not
only the common cases.

EDA and Data Cleaning

EDA

Analyzing datasets to summarize their notable characteristics is called

Exploratory Data Analysis.
Helps in performing investigations on data so as to discover hidden patterns,
anomalies etc.
Aids in checking assumptions and hypothesis with the help of summary statistics.

Data Cleaning

Data can have several shortcomings. To list a few are

Missing values
Duplicate data
Invalid data

The process of detecting, correcting and ensuring that the given dataset is
error free, consistent enough to use is called Data Cleaning.

Choosing a Model

There are numerous models that researchers and Data scientists have created over
the years.
Some are very well-suited for image data, while others are suited for sequences,
text-based data and many more.
Choosing the right model for the problem will impact the efficiency of the
model.
Training

The next step of the Machine Learning process, often known as the the bulk of ML
is Training the model.
This step is very similar to a person who is learning to drive for the first
time. Though at first they dont know any of pedals, switches, breaks but eventually
after lots of practice and feedbacks a licensed driver emerges.
The data is split into Training Data and Testing Data.
Model is trained with the training data using different ML algorithms by
adjusting the parameters in multiple iterations.
Testing Data are put aside as unseen data to evaluate your models.

Evaluation

Once training is complete, it’s time to see if the model is any good, using
Evaluation.
This is where that dataset that we set aside earlier comes into play(i.e)
Testing Data.
Evaluation allows us to test our model against data that has never been used for
training.
This metric allows us to see how the model might perform against data that it
has not yet seen.
This is meant to be representative of how the model might perform in the real
world.

Hyperparameter tuning

After the evaluation step, it's time to see if we can improve our training
furthermore by tuning different parameters that were implicitly assumed in the
training process and this process is called Hyperparameter Tuning.
The tuned model is once again evaluated for model performance, and this cycle
continues until the final best performing model is chosen.

Interpret and Communicate

The most challenging task of the ML project is explaining the model's output.
Earlier days, Machine learning is considered to be a BlackBox because it was
hard to interpret their insights and values.
The more interpretable your model is, then more it is easier to communicate your
model's importance to the stakeholders

Deployment and Documentation

Model deployment often poses a problem because of the coding and data science
experience it requires and because the time-to-implementation of traditional data
science methods from the start of the cycle is prohibitively long.
The trained model has to be deployed in a real-world system for it to be
efficient to humans.
It can be deployed using any of the frameworks like FLASK, Cloud, Azure etc.
Document your project well for your descendants to handle it.

Types of Machine Learning

The types of Machine Learning are as follows:

Supervised Learning
Unsupervised Learning
Reinforcement Learning

Supervised Learning

Supervised learning is the machine learning task of learning a function that

maps an input to an output based on example input-output pairs.
It infers a function from labeled training data.
Each training example is a pair consisting of an input object and a desired
output value.
A supervised learning algorithm analyzes the training data and produces an
inferred function, which can be used for mapping new examples.
Applications

Spam Detection
Pattern Recognition
Speech Recognition

Unsupervised Learning

Unsupervised Learning helps in uncovering hidden patterns from unlabeled data.

Applications

Recommender Systems
Targetted Marketing
Customer Segmentation
Structure Discovery

Reinforcement Learning

Reinforcement Learning is a type of machine learning in which software agents ought

to take actions in an environment so as to maximize some notion of cumulative
reward.

Applications

Genetics
Economics
Robot Navigation

The image depicted above illustrates how to integrate the process of Machine
Learning into the traditional Software Development Life Cycle (SDLC).

The three phases include:

Planning
Data Engineering
Modeling

Machine Learning Terminologies

Accuracy

Accuracy is the percentage of correct predictions made by the model.

Algorithm

Machine learning algorithms are programs (math and logic) that adjust themselves
to perform better as they are exposed to more data.
The learning part of machine learning means that those programs change how they
process data over time, much as humans change how they process data by learning.
 So a machine-learning algorithm is a program with a specific way to adjusting
its own parameters, given feedback on its previous performance making predictions
about a dataset.

Examples

Linear regression
Decision trees
Support vector machines
Neural networks

Machine Learning Terminologies

Categorical Variables

Categorical variables are variables with a discrete set of possible values.

They can be ordinal or nominal.

Classification

Classification aids in predicting the categorical output.

Clustering

Clustering is the unsupervised grouping of data into buckets.

Machine Learning Terminologies

Dimension

The dimension of data denotes the number of features in a dataset.

Feature

For a dataset, a feature represents the combination of attribute and value.

Feature Selection

Feature selection is the process of selecting relevant features from a dataset for
creating a Machine Learning model.
Machine Learning Terminologies

Hyperparameters

Hyperparameters are higher-level properties of a model, such as how fast it can

learn or the complexity of a model.

Instance

An instance is a data point, row, or sample in a dataset.

Label

The label is the answer part of the observation in supervised learning.

Machine Learning Terminologies

Outlier

An outlier is an observation that deviates significantly from other observations in

the dataset.

Regression

Regression predicts the continuous form of output (For example, price, sales, and so
on).

Validation Set

The validation set is a set of observations used during model training to provide
feedback on how well the current parameters generalize beyond the training set.
Let us now explore the following popular Machine Learning techniques:

Classification
Clustering
Association Rule Mining
Outlier Detection
Regression

Classification
Definition

Classification is the process of identifying a category to which a new

observation belongs, based on a training set of data containing observations whose
categories are already known.

It follows a two-step process, namely:

Learning Step - Training phase where a model is constructed.
Classification Step - Predicting the class labels and testing the same for
accuracy.
Classification predicts the value of the categorical variables.

Clustering

Clustering is the task of grouping a set of objects, such that objects in the
same cluster are similar to each other when compared to the objects in the other
clusters.

Distance measure plays a significant role in clustering.

Clustering is an unsupervised learning method.

 The common distance measures used in various datasets are as follows.

Numeric Dataset

- Manhattan distance
- Minkowski distance
- Hamming distance

Non-Numeric Dataset

- Jaccard index
- Cosine Similarity
- Dice Coefficient

Association Rule Mining aids in identifying the associations, correlations, and

frequent patterns in data.

The derived relationships are represented in the form of Association Rules.

Outlier Detection

Jiawei Han defines Outlier as

A data object that deviates significantly from the normal objectsas if it were
generated by a different mechanism.

The types of Outlier are as follows:

Global Outlier

Global Outlier significantly deviates from the entire dataset.

Contextual Outlier

Contextual Outlier significantly deviates based on the context selected.

Collective Outlier

Collective Outlier is a subset of data objects that collectively deviates from

the entire dataset.

Regression

Regression analysis is a statistical method that aids in examining the

relationship between two or more variables of interest.
Examines the influence of one or more independent variables on a dependent
variable.

Decision Tree
 A Decision Tree (DT) is a tree-like model of decisions and possible
consequences, chance event outcomes, resource costs, and utility.
Decision Trees are a non-parametric supervised learning method used for
classification and regression.
Naive Bayes

A Naive Bayes classifier is a probabilistic Machine Learning model that is used for
classification tasks. The crux of the classifier is based on the following Bayes
theorem formula.

P(A∣B)=P(B∣A)P(A)P(B)P(A|B)=\dfrac{P(B|A)P(A)}{P(B)}P(A∣B)=P(B)P(B∣A)P(A)

Support Vector Machine

Support Vector Machine (SVM) is a supervised machine learning algorithm. It is used

for classification or regression type of problems.
K-means Clustering

Delve into this video to know about a type of clustering algorithm called K-means
Clustering.
Random Forest
Linear Regression
Logistic Regression
Deep Learning is a subset of Machine Learning, which in turn, is a subset of
Artificial Intelligence.

Artificial Intelligence

Artificial Intelligence (AI) is a technique that helps machines to mimic human

behavior.

Machine Learning

Machine Learning is an application of AI that allows the system to learn and improve
from experience automatically.

Deep Learning

Deep Learning is a type of Machine Learning that is inspired by the structure of the
brain. It is also known as Artificial Neural Network (ANN). It uses complex
algorithms and deep neural networks to train models.
What is Deep Learning?

Definition

Deep Learning involves networks which are capable of learning from data and
functions similar to the human brain.
Why Deep Learning?

Let us explore the reasons which make Deep Learning shine in the industry.

Processes massive amount of data

Deep Learning can process an enormous amount of both Structured and Unstructured
data.

Performs Complex Operations

Deep Learning algorithms are capable enough to perform complex operations when
compared to the Machine Learning algorithms.
Why Deep Learning?

Achieves Best Performance

As the amount of data increases, the performance of Machine Learning
algorithms decreases.
On the other hand, Deep Learning maintains the performance of the model.

Feature Extraction
Machine Learning algorithms extract patterns from labeled sample data, while
Deep Learning algorithms take large volumes of data as input, analyze them to
extract the features on its own.

Machine Learning
If done through Machine Learning, we need to specify the features based on which the
two can be differentiated like size and stem, in this case.

Deep Learning

In Deep Learning, the features are picked by the Neural Network without any human
intervention. But, that kind of independence can be achieved by a higher volume of
data in training the machine.
Neural Networks
Neural Networks

The human brain contains billions of cells called Neurons. The structure of a
neuron is depicted in the above image.
Neural Networks is a set of algorithms designed to learn the way our brain
works.
The biological neurons inspire the structure and the functions of the neural
networks.

Biological and Artificial Neurons - Terminologies

Biological Neuron Artificial Neuron
Dendrites Inputs
Nucleus Nodes
Synapse Weights
Axon Output
A Node is also called a Neuron or Perceptron.
The basic structure of an Artificial Neural Network (ANN) consists of artificial
neuron that are grouped into 3 different layers namely:

There are three different layers in a neural network, namely:

Input Layer
Hidden Layer
Output Layer

Input Layer

The input layer communicates with the external environment and presents a
pattern to the neural network.
Every neuron in the input layer represents an independent variable that
influences the output.

Structure of an Artificial Neural Network

Hidden Layer

A Neural Network consists of several hidden layers, each consisting of a

collection of neurons.
The hidden layer is an intermediate layer found between the input layer and the
output layer.
 This layer is responsible forextracting the features required from the input.
There is no exact formula for calculating the number of the hidden layers as
well as the number of neurons in each hidden layer.

Output Layer

The output layer of the neural network collects and transmits information in the
desired format.

Every single layer is made up of one or more number of neurons. A simple model of an
Artificial neuron is shown in the figure.

Parameters represented are:

X1,X2,...,XmX_1,X_2,...,X_mX1,X2,...,Xm are the input signals.

W1,W2,...WmW_1,W_2,...W_mW1,W2,...Wm are the weights.
∑\sum∑ is the Summation function
σ\sigmaσ is the Activation function.
bkb_kbk represents the bias value

Artificial Neural Network - Process Flow

The process flow depicted in the image is explained as follows:

A node is where the computations happen

A node combines the Input data X1,X2,...,XmX_1,X_2,...,X_mX1,X2,...,Xm with a
set of Weights W1,W2,...,WmW_1,W_2,...,W_mW1,W2,...,Wm which in turn aids in
amplifying or diminishing the input with respect to the polarity attached to it.
On multiplying the inputs with weights, a specific significance is assigned to
the inputs with regard to the task the algorithm is learning. It helps in answering
the question Which input is the most helpful in learning data without error?
The input-weight products are summed up using the Summing function, and the
output yielded is passed to the Activation function to decide whether and to what
extent the signal should progress to have an effect in the final output.

Summation and Activation Functions

Summation Function

Various inputs are multiplied with their respective connection weights and
summed up together with the bias value.

∑j=1NXjWj+bk\sum_{j=1}^{N}X_jW_j+b_k∑j=1NXjWj+bk

where:

XjX_jXj represents the inputs

WjW_jWj represents the weights

Summation and Activation Functions
Activation Function

Activation Function aids in deriving the output.

It is also known as the Transfer Function.

It maps the resulting values between 0 to 1 or -1 to 1.

There are two types of Activation function, namely:

Linear Activation Function
Non-Linear Activation Function

Non-Linear Activation Function is the most commonly used Activation function in

Neural Networks.

The default Activation Function is ReLU.

In January 2018, Google's CEO Sundar Pichai claimed that Artificial intelligence
(AI) would be more transformative to humanity than electricity.
Learning Process of a Neural Network

The learning process of a Neural Network includes updating the network architecture
and connecting the weights for the network to perform efficiently.

Designing a Learning Process involves the following:

Learning Paradigm

Having a model from the environment in which Neural Network works.

Learning Rules

Figuring out the rules that aid in updating the weights.

Learning Algorithms

Identifying the procedure to update the weights according to the learning rules.
Learning Paradigm

The following are the various Learning Paradigms in Neural Networks:

Supervised
Unsupervised
Reinforcement

Learning Rules

The four basic types of Learning Rules in Neural Network are:

Error Correction Rules

 Hebbian
Boltzmann
Competitive Learning

Learning Algorithms

Following are a few Deep Learning algorithms:

Adaptive Resonance Theory

Kochen Self Organization Map
ADALINE
Perceptron
Backpropagation

Principles

A Neural Network works based on two principles, namely:

Forward Propagation
Backward Propagation

Neural Network Architecture

The architecture of a Neural Network can be broadly classified into two, namely:

Feed Forward Artificial Neural Network

Recurrent Neural Network

Feed Forward Artificial Neural Network

Feed Forward Artificial Neural Network

The information must flow from input to output only in one direction.
No Feedback loops must be present.

A few Feed Forward Artificial Neural Networks are:

Single-layer Feed Forward Network

Multi-layer Feed Forward Network

Recurrent Neural Network

Recurrent Neural Network

Information can be transmitted in both directions.

Feedback loops are allowed.

A few Recurrent Neural Networks are:

Hopfield Network
Elman Network
 Jordan Network

MELTANT-α\alphaα is an avatar robot with the world’s most powerful and flexible
robotic hand which directly imitates human hand movements.
Neural Networks are employed in various fields. Following are a few types of Neural
Networks that we will explore in this section:

Radial Basis Function

Long/Short Term Memory
Gated Recurrent Unit
Autoencoder
Convolutional Neural Networks

Radial Basis Function

Radial Basis Function is similar to Feed Forward Network but uses Logistic function
as the Activation Function.
long/Short Term Memory

Long/Short Term Memory (LSTM) networks are used to classify, process, and make
predictions based on time series data.
LSTM networks can predict the action in a specific video frame by keeping in
mind the action that occurred in the earlier frames.
The applications of LSTM include writing, speech recognition, and so on.

Gated Recurrent Unit

Gated Recurrent Unit (GRU) is LSTM with a forget gate.

It is used in sound, speech synthesis, and so on.

Autoencoder

Autoencoder (AE) is used for classification, clustering, and feature

compression.
It is trained without supervision.

Image Classification

Image classification is one of the common applications of deep learning.

A convolutional neural network can be used to recognize images and label them
automatically.

Object Recognition
Object Recognition

Deepnets (an optimized version of Deep Neural Networks) can be trained to

recognize different objects within the same image.

Video Recognition
Video Recognition

Deepnets can also be trained for real-time object recognition with respect to
time and space.

Some of the famous use cases are driverless cars, robots, and theft detection.

Sentiment Analysis

Using sentiment analysis, the underlying intent of the text can be extracted.

With social media channels, it is possible to automate and measure the feelings
of the public on a given news story, topic, brand, or product.

Positive sentiment can be identified, thereby allowing the marketing of a

product, or understanding which elements of a business strategy are working.

Medical Applications

Deepnets can be trained to detect cancerous cells, benign and malignant tumors
from MRI and CT scans.

They are also applied in drug discovery by training nets with molecular
structure and chemical compositions.

Deepnet Platform

A deepnet platform is a service that allows you to incorporate deepnet in your

applications without building one from scratch.
This platform provides a set of tools and interfaces to create a custom deepnet.

Deepnet platforms are of two types, namely:

Software platform: This platform is available as downloadable packages that need

to be deployed on your hardware.
Full platform: It is available as an online interactive UI to build and deploy
models without any coding experience.

Deepnet Platforms - Tools

The following are the tools offered by the deepnet platforms:

Deepnet capability
Data Munging
UI/Model Management
Infrastructure

H2O.ai
H2O.ai