TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
Introduction to ml
1. Introduction to ML with a Simple Demo
Dr.Girija Narasimhan 1
Dr. Girija Narasimhan
University of Technology and Applied Sciences
IT Department,
Muscat, Sultanate of Oman
2. Dr.Girija Narasimhan 2
The goal of machine learning generally is to understand the
structure of data and fit that data into models that can be
understood and utilized by people.
GOAL
3. Dr.Girija Narasimhan 3
.
AI VS. ML VS. DL
Artificial Super Intelligence (ASI)
Artificial Narrow Intelligence (ANI)
Artificial General Intelligence (AGI)
Classification of AI
4. Dr.Girija Narasimhan 4
AGI model is precisely defined as it will act appropriately like human-level AI.
The other terms used to scientifically describe AGI model obtain “computational
intelligence”, “natural intelligence”, “cognitive architecture”, “biologically inspired cognitive
architecture” (BICA).
General Intelligence means it includes the ability to attain a variety of goals, and carry out a
variety of tasks, in a variety of different circumstances and surroundings.
It means how human beings are applying knowledge in diverse fields for example, Like Cho,
SPB.
The combination of AGI and machine learning can themselves recognize and acquire any
type of intellectual task. This combination is called Strong AI.
Artificial General Intelligence (AGI)
5. Dr.Girija Narasimhan 5
Core AGI hypothesis: the creation and study of synthetic
intelligence with sufficiently broad (e.g. human-level) scope
and strong generalization capability are at the bottom
qualitatively different from the creation and study of synthetic
intelligence with a significantly narrower scope and weaker
generalization capability.
AGI hypothesis
6. Dr.Girija Narasimhan 6
• AGI models are still under empirical research.
• The AGI researchers are focusing on this efficient model which is continuously undergoing the various
level of cognitive tests to attain human Intelligence.
• The first ultimate test attend the Loebner Prize competition.
• This prestigious competition is an annual competition in artificial intelligence. The key aspect of the
Loebner Prize competition to promptly check the standard of turning the test of the model.
• The Intelligence level of audio-video conversation and responses.
• AGI coffee test. For the coffee test, AGI enters any kitchen and finds the ingredients like sugar, water,
coffee powder and finding coffee cub and perfectly mix all.
• Robot College Student test, employment test.
7. Dr.Girija Narasimhan 7
AI is a bigger concept to create
intelligent machines that can
simulate human thinking capability
and behavior
Differentiate AI and ML
According to Arthur Samuel, Machine
learning is an application or subset
of AI that allows machines to learn from
data without being programmed explicitly
AI is decision making
ML allows system to learn new things
from data
8. Dr.Girija Narasimhan 8
Deep learning is a subset of ML
In other words, DL is the next evolution of machine learning.
DL algorithms are roughly inspired by the information processing patterns found
in the human brain.
Whenever brain receive a new information, the brain tries to compare it to a
known item before making sense of it — which is the same concept deep
learning algorithms employ.
Deep Learning
10. Dr.Girija Narasimhan 10
Machine learning is to build algorithms that can receive input data
Machine learning (ML) is a category of an algorithm that allows software applications to
become more accurate in predicting outcomes
use statistical analysis to predict an output
How predicting
ML Process
11. 11
Machine learning can also be used in the prediction systems. Considering the loan example,
to compute the probability of a fault, the system will need to classify the available data in
groups.
Prediction [Ref. 4]
Lenddo – Digital Footprint Analysis
Company started in 2011, claim 5 million people receiving loans via their partners
because their system was able to evaluate their creditworthiness.
ZestFinance – Artificial Intelligence and Search-Based Analysis
Using machine learning to process alternative data to get information on so-called
“thin file borrowers” — those with no or little credit history
Equifax
Using machine learning to better determine an individual’s creditworthiness
Streamlining
Eliminating administrative overhead and delays is a way to maximize the
amount of profits for each loan created
Increase business and gain customers
Upstart – Full Automation and AI Determined Creditworthiness
Amazon – Small Business Loans
12. Dr.Girija Narasimhan 12
Machine learning can be used for face detection in an image as well. There is a separate category for each
person in a database of several people.
classification for face
detection methods
Image recognition [Ref.5]
13. Dr.Girija Narasimhan 13
1.Knowledge-Based:-
The knowledge-based method depends on the set of rules, and it is based on human knowledge to detect the
faces. Ex- A face must have a nose, eyes, and mouth within certain distances and positions with each other.
The big problem with these methods is the difficulty in building an appropriate set of rules.
2.Feature-Based:-
The feature-based method is to locate faces by extracting structural features of the face. It is first trained as a
classifier and then used to differentiate between facial and non-facial regions. The idea is to overcome the limits of
our instinctive knowledge of faces. This approach divided into several steps and even photos with many faces
they report a success rate of 94%.
3.Template Matching:-
Template Matching method uses pre-defined or
parameterized face templates to locate or detect the
faces by the correlation between the templates and
input images. Ex- a human face can be divided into
eyes, face contour, nose, and mouth.
14. Dr.Girija Narasimhan 14
4.Appearance-Based:-
The appearance-based method depends on a set of delegate training face images to find out face models. The
appearance-based approach is better than other ways of performance. In general appearance-based method rely on
techniques from statistical analysis and machine learning to find the relevant characteristics of face images.
The appearance-based model further divided into sub-methods for the use of face detection which are as follows-
4.1.Eigenface-Based:-
Eigenface based algorithm used for Face Recognition, and it is a method for efficiently representing faces using
Principal Component Analysis.
4.2.Distribution-Based:-
The algorithms like PCA and Fisher’s Discriminant can be used to define the subspace representing facial patterns.
There is a trained classifier, which correctly identifies instances of the target pattern class from the background
image patterns.
4.3.Neural-Networks:-
Many detection problems like object detection, face detection, emotion detection, and face recognition, etc. have
been faced successfully by Neural Networks.
4.4.Support Vector Machine:-
Support Vector Machines are linear classifiers that maximize the margin between the decision hyperplane and the
examples in the training set. Osuna et al. first applied this classifier to face detection.
15. Dr.Girija Narasimhan 15
4.5.Sparse Network of Winnows:-
They defined a sparse network of two linear units or target nodes; one represents face patterns and other
for the non-face patterns. It is less time consuming and efficient.
4.6.Naive Bayes Classifiers:-
They computed the probability of a face to be present in the picture by counting the frequency of
occurrence of a series of the pattern over the training images. The classifier captured the joint statistics of
local appearance and position of the faces.
4.7.Hidden Markov Model:-
The states of the model would be the facial features, which usually described as strips of pixels. HMM’s
commonly used along with other methods to build detection algorithms.
4.8.Information Theoretical Approach:-
Markov Random Fields (MRF) can use for face pattern and correlated features. The Markov process
maximizes the discrimination between classes using Kullback-Leibler divergence. Therefore this method
can be used in Face Detection.
4.9.Inductive Learning:-
This approach has been used to detect faces. Algorithms like Quinlan’s C4.5 or Mitchell’s FIND-S used
for this purpose.
16. Dr.Girija Narasimhan 16
FINANCE [REF.7]
ZESTFINANCE
ZestFinance is the maker of the Zest Automated Machine Learning (ZAML) platform, an AI-powered
underwriting solution that helps companies assess borrowers with little to no credit information or
history.
DataRobot
DataRobot helps financial institutions and businesses quickly build accurate predictive models that
enhance decision making around issues like fraudulent credit card transactions, digital wealth
management, direct marketing, Blockchain, lending and more.
It provides machine learning software for data scientists, business analysts, software engineers,
executives and IT professionals.
SCIENAPTIC SYSTEMS
Scienaptic Systems provides an underwriting platform that gives banks and credit institutions more
transparency while cutting losses. Scienaptic boasted $151 million in loss savings in just three weeks.
17. Dr.Girija Narasimhan 17
Underwrite.ai
Underwrite.ai analyzes thousands of data points from credit bureau sources to assess
credit risk for consumer and small business loan applicants.
The platform acquires portfolio data and applies machine learning to find patterns and
determine good and bad applications. Because of its accuracy, Underwriter.ai claims it
can reduce defaults by 25-50%.
KENSHO, AYASDI, ALPHASENSE
18. Dr.Girija Narasimhan 18
Current Applications of AI in Medical Diagnostics
Many of today’s machine learning diagnostic applications appear to fall under the following
categories:
Oncology: Researchers are using deep learning to train algorithms to recognize cancerous
tissue at a level comparable to trained physicians.
Pathology: Pathology is the medical specialty that is concerned with the diagnosis of
disease based on the laboratory analysis of bodily fluids such as blood and urine, as well as
tissues. Machine vision and other machine learning technologies can enhance the efforts
traditionally left only to pathologists with microscopes.
Rare Diseases: Facial recognition software is being combined with machine learning to help
clinicians diagnose rare diseases. Patient photos are analyzed using facial analysis and deep
learning to detect phenotypes that correlate with rare genetic diseases.
Medical diagnoses [Ref.6]
19. Dr.Girija Narasimhan 19
Chatbots: Companies are using AI-
chatbots with speech recognition
capability to identify patterns in patient
symptoms to form a potential diagnosis,
prevent disease and/or recommend an
appropriate course of action.
20. Dr.Girija Narasimhan 20
Amazon Echo Dot
Google Home Mini.
Apple HomePod.
Azatom Venture Smart Speaker.
Zolo Halo Smart Speaker.
Sonos One.
Top Virtual Assistant in Market
It is the translation of spoken words into the text. It is used in voice searches and more. Voice user interfaces
(VUI) include voice dialing, call routing, and appliance control. It can also be used a simple data entry and the
preparation of structured documents. voice command device (VCD) is a device controlled with a voice user interface.
Automatic speech recognition (ASR)
21. Dr.Girija Narasimhan 21
Google wants its AI-powered voice assistant to spread to every corner of tech
Google AI
Google Nest, previously named Google Home, is a line of smart speakers developed
by Google under the Google Nest brand.
The devices enable users to speak voice commands to interact with services
through Google Assistant, the company's virtual assistant.
22. Dr.Girija Narasimhan 22
Types of Machine Learning
Machine learning can be classified into 3 types of algorithms.
1.Supervised Learning
2.Unsupervised Learning
3.Reinforcement Learning
24. Dr.Girija Narasimhan 24
Supervised Learning Algorithm
In Supervised learning, an AI system is presented with data which is labeled, which means that each data
tagged with the correct label.
‘Spam’ or ‘Not Spam’. This labeled data is used by the training supervised model, this data is used
to train the model.
25. Dr.Girija Narasimhan 25
Unsupervised Learning Algorithm
In unsupervised learning, an AI system is presented with unlabeled, uncategorized data and
the system’s algorithms act on the data without prior training. The output is dependent upon
the coded algorithms. Subjecting a system to unsupervised learning is one way of testing AI.
In our training data, we don’t provide any label to the
corresponding data. The unsupervised model is able to
separate both the characters by looking at the type of data
and models the underlying structure or distribution in the
data in order to learn more about it.
some characters to our model which
are ‘Ducks’ and ‘Not Ducks’.
26. Dr.Girija Narasimhan 26
Python leads the pack, with 57% of data scientists and machine learning developers using it
and 33% prioritizing it for development.
Given all the evolution in the deep learning Python frameworks over the past 2 years,
including the release of TensorFlow and a wide selection of other libraries.
Both TensorFlow and PyTorch have their advantages as starting platforms to get into neural
network programming.
Python
Programming Languages
29. Dr.Girija Narasimhan 29
10 Best Machine Learning Certification for 2020 [Ref.9]
1. Professional Certificate Program in Machine Learning and Artificial Intelligence
2. Machine Learning with TensorFlow on Google Cloud Platform Specialization
3. Machine Learning Stanford Online
4. Professional Certificate in Foundations Of Data Science
5. Certification of Professional Achievement in Data Sciences
6. eCornell Machine Learning Certificate
7. Certificate in Machine learning
8. Harvard University Machine Learning
9. Machine Learning with Python
Google's Teachable Machine Uses TensorFlow.js to
Bring Code-Free Machine Learning to the Browser
30. Dr.Girija Narasimhan 30
The standard – ITU Y.3172 – describes an architectural framework for networks to
accommodate current as well as future use cases of Machine Learning.
ISO/IEC CD 23053.2
Framework for Artificial Intelligence (AI) Systems Using Machine Learning
(ML)
Machine Learning Standard
32. Dr.Girija Narasimhan 32
Google Creative Lab
Tom Seymour is an award winning creative lead, working at the Google Creative Lab in London
Skills
3D Design, Advertising, Art Direction, Brain storming, Brand / Logo Design, Communications, Creative
Direction, Digital Art, Digital Marketing, Digital Strategy, Directing, Entrepreneurship, Exhibition Design,
Experiential Marketing, Film, Furniture Design, Graphic Design, Interactive Design, Interface Design, Int
erior Design, Lighting Design, Photography, Photoshop, Problem Solving, Script Writing, UX/UI, Web D
esign
https://medium.com/@techandthecity/inside-google-creative-lab-5f148b0e8f3c
33. Dr.Girija Narasimhan 33
It is a “small team within Google”, as Seymour put it.
They work for any Google projects — from Android to Chrome. the whole point of the
team is to communicate what Google has to offer. Yes, it is an advertisement, but as well
as Google is not just a search engine, the Creative Lab is not an ordinary advertisement
department.
The team is the mix of people with background from design, fashion, filmmaking.
Their key principles of Google Creative Lab:
1.Know the User
2. Know the MAGIC (the essence and all the details of the particular Google’s project )
3. Connect the two
Google Creative Lab
35. Dr.Girija Narasimhan 35
Computer Vision is a type of Artificial Intelligence (or AI) where people train a computer to
recognize objects.
Computer with internet connection and webcam (you can also use your phone!)
Computer Vision
36. Dr.Girija Narasimhan 36
"Objectifier-Spacial Programming" video in order to get a glimpse at the future of algorithm customization.
This includes teaching the algorithm to :
turn the light one when you open a book
turn the light off when you lie on bed
stop the music or start the music with gestures
Objectifier Spatial Programming (OSP) empowers people to train objects in their daily environment to respond
to their unique behaviors.
It gives an experience of training an artificial intelligence; Train objects in your environment to respond to your
behavior
Objectifier-Spacial Programming
37. Dr.Girija Narasimhan 37
https://www.youtube.com/watch?v=DFBbSTvtpy4
https://www.youtube.com/watch?v=kwcillcWOg0
Introducing Teachable Machine 2.0 (from Google Creative Lab)!
Train a computer to recognize your own images, sounds, & poses
A fast, easy way to create machine learning models – no coding required.
You can download your model or host it online for free.
The Coding Train
https://experiments.withgoogle.com/teachable-machine
Teachable Machine Tutorials
Teachable Machine 2.0
https://teachablemachine.withgoogle.com/v1/
https://design.google/library/designing-and-learning-teachable-machine/
PAIR: the People + AI
Research Initiative
Team
38. Dr.Girija Narasimhan 38
All training is done in the browser using the deeplearn.js library.
It is a hardware-accelerated JavaScript library built by the Google Brain PAIR team that is freely
available.
The library was announced in August 2017 on the Google blog, and several applications that use
the library are available on the deeplearn.js website.
Teachable Machine 2.0 Library
39. Dr.Girija Narasimhan 39
Teachable Snake is an interactive web game powered by the beta version
of Teachable Machine 2 and React. js, inspired by Webcam Pacman project.
Teachable Snake
41. Dr.Girija Narasimhan 41
Step 2: Creating New Project
Save model as image and audio and pose
Step 3: Creating New Project
click
Step 4: Edit sample title- Face and water bottle
44. Dr.Girija Narasimhan 44
Step 7: Downloaded Face-samples available as Zip format in the download folder
Step 8: unzip the fact-samples.zip – you can find three samples 0.jpg, 1.jpg, 2.jpg