Innovation and Startup Scope in Artificial Intelligence

Dr. Thyagaraju G S

Professor and HoD , Dept of CSE

SDM Institute of Technology – Ujire

Acknowledgments : I would like to thank all researchers , scientists , engineers ,innovators, industries , research centers and institutions around the world whose content is being used in this presentation , purely for academic purpose.

Agenda

​

1. What ? Why ? Who ? : AI.

2. Scope for Innovation in AI

3. Scope for Startup in AI

​

What is AI?

Artificial Intelligence is a field of study to design and develop software programs for machines that perform human like cognitive tasks without explicit programming.

Cognition refers to a range of mental processes relating to the acquisition, storage, manipulation, and retrieval of information.

What is AI?

• Imagination
• Empathy
• Curiosity
• Resilience
• Creativity

• Emotional Inteligence
• Teaming
• Social Intelligence
• Sense Making
• Critical Thinking
• Adaptive Tinking

Human capabilities

Innate but

can be amplifeid

Developed through

Experience and

Practice

Why AI?

Why AI

Human beings are being unable to absorb, interpret and make complex decisions based on the data .

​

Work demand is growing in terms of volume , complexity , repetitive and risk.

Availability of voluminous data , better algorithms ,

high performing storage and computing systems.

To predict and discover the hidden insights in the voluminous pattern data.

Ai for Who?

Scope for Innovation in AI

• Enabling objects (Any) with
• Human Cognitive Capabilities [ Act like a human being]
• Contextual Intelligence [ Act Better than human]
• Prediction Capabilities [ Like predicting next product]
• Diagnostic Capabilities [ Self Maintenance ]

​

​

​

*Objects : Software / Hardware / Any tangible / intangible things / Living /Non Living /Anything

Scope for Innovating Every day Objects

Ceiling Fan

• Existing Capabilities

• Innovative Capabilities

Every Day Objects

Every Day Objects

Abilities of AI Object

Intelligent Object

Imagine any Object and Intellectual abilities that object can possess .

Example 1 : Intelligent Pen

• Touch Sensitive , Have Memory , Captures and recognizes Handwriting Characters / strokes
• Context Aware (self ,owner and environment)
• Learn user handwritten pattern and recognize
• Transforms the Handwritten Characters to Digital Data
• Uploads the data into cloud or computer
• Records voice
• Recognize the user
• Interact with user and other objects
• Provide notifications if it is lost /in the other hands/expired.

​

Example 2 : Intelligent Mobile Phone

• Context Aware [ Location , Time , User Situation ]
• Learn the user usage Pattern
• Context based service recommendation with minimal user interaction
• Natural interaction with user and other objects
• Personal Assistant , Dairy and Planner
• Cloud Data Storing : Importance based
• Self Cleaning and updating with minimal notifications

​

Example 3: Intelligent Television

• Context Aware
• Social Aware
• User Context based service recommendation
• Gesture/Voice/Text based natural interaction with user and objects
• Ability to learn about user and environment

Example 4: Intelligent Classroom

• Context sensitive settings of seating arrangement , Temperature , lighting , healthy measures and clean maintenance .
• Recognize and interact with stake holders ( Students and Teachers)
• Student intellectual (IQ ,EQ,SQ) context aware ubiquitous teaching , learning , evaluation and progress reporting .
• Teachers assistive apps to facilitate Learning through Hands on sessions , Quiz , Demonstrations, Use cases , Models , Prototypes , Live Interactions, Videos , Audios , E- Books and online courses .
• Discussion Forum to interact with peers and teachers.
• Teacher and Subject centric teaching material development and quality teaching maintenance tools on the shelf ( E-books , Video Recording ,etc).
• Automated Work Planner and Execution Dairy
• Apps for Exam Conduction , Evaluation and Result Analysis.
• Students related data entry and data analysis should be automated.

​

AI Vs ML Vs DL

Artificial Intelligence ��Deduction , �Reasoning , �Knowledge Representation ,�Expert Systems , �Neural Network , �Planning ,�Robotics , �Computer Vision ,�Natural Language Processing , �Machine Learning ��3 Domains of AI :�Data �Computer Vision �Natural Language Processing

ML Techniques

1. Supervised Learning
• Classification:

1. Find S Algorithm 2. Candidate Elimination Algorithm 3. Decision Tree Algorithm 4. Naïve Bayes Classifier 5. Support Vector Machine 6. K NN Algorithm 7. Logistic Regression

• Regression:

1. Linear Regression 2. Locally weighted Regression 3. Multi Linear Regression 4. Polynomial Regression 5. LASSO Regression

1. Unsupervised Learning :

1. Clustering: K - Means , EM Algorithm 2. Dimensionality Reduction: PCA 3. Association Analysis: FP Growth

1. Reinforcement Learning :

a. Q – Learning b. Genetic Algorithm

DL Techniques

• Classic Neural Networks.
• Convolutional Neural Networks. ...
• Recurrent Neural Networks (RNNs) ...
• Generative Adversarial Networks. ...
• Self-Organizing Maps. ...
• Boltzmann Machines. ...
• Deep Reinforcement Learning. ...
• Autoencoders.
• Backpropagation
• Gradient descent

Design Thinking for Artificial Intelligence

There is no universally accepted approach about the implementation of AI. It is necessary to study the current AI model and implement design thinking into it.

​

Different Phases of Design Thinking for AI

Problem Scoping

1. Empathize
2. Define

Modeling

1. Ideate
2. Data Preparation
3. Identify the AI capabilities needed: ML/NLP/Expert Systems /Vision /Speech
4. N Models for each Idea

Prototyping

1. Model Tuning
2. Agree on the right SDLC model for the project.[ Water Fall Model : RA, Design , Development , Testing and Deployment]
3. Prototype Building
4. Testing and Finalizing

Deployment

Problem Statement

Best Idea and Model

Prototype

Use Case 1: Intelligent Fan (Conceptual Design)

Problem Scoping :Empathy

• Fan is not context aware .
• Fan is not recognizing the user and his/her situation.
• Fan should be switched on or off manually.
• Fan speed should be set /alter manually .
• Fan does not know how to adjust the Speed by itself as per user requirement.
• Fan is not aware about the speed and temperature relationship.
• Fan can not understand the natural language. (cannot talk)
• Fan is not aware about the electricity bill. (power consumption).
• Fan is not aware about its own purpose .
• Fan does not remember the past history .
• Fan does not know who are its users.
• Fan does not clean itself

​

Problem Scope : Define

• Conceptual design of intelligent Fan which can provide context sensitive services.

Modeling : Ideate

• Idea1 : Intelligent Fan which can recognize the user's context and sets its speed based on the set of rules.
• Idea2 : Intelligent Fan which learns on its own to provide contextual service based on history.

Modeling : Data Preparation

• Rule Base
• Users Image Data Set
• Instructions sets
• Voice commands
• Pairable Devices

​

​

Modeling : AI capabilities

• Recognizing user and context
• Decision Making
• Recommendation
• Context based service recommendation
• Voice Communication

​

​

Prototype (Conceptual )

Use Case 2: Intelligent Mobile

• It would be more comfortable for the user, if the mobile phone present in the meeting room can provide meaningful services (like going to silent mode automatically as soon as the meeting starts , allowing only emergency calls, answering to known and unknown calls suitably, linking to meeting related sites, etc., without interrupting the meeting session), based on context.
• Mobile phone should recognize its location and behave accordingly. For example , suppose if student carries mobile into classroom , mobile must be aware about the class and policies and should go into classroom mode. When mobile is in classroom mode , it must behave as a learning gadget………….

Scenario2

Scenario1

AI Abilities : Talk , Listen , Recognize user context, Decide , Recommend, Understanding User Satisfaction , Notification, Answering Call , ..

Mobile Rule Base

Bayesian Probability based Learning

Bayesian Probability based Learning

Overall Performance of CAMP Recommendation System

Use Case3 : Intelligent Television

AI Abilities :

Talk , Listen , Recognize user context , Decide , Recommend, understand User Satisfaction , Notification , ……..

Thank You All