Spotle.ai Study Material
Spotle.ai/Learn
The Power Of Graph Computing For Deep Learning
Introduction to
Spotle.ai Study Material
Spotle.ai/Learn 2
Instructors
Mousum Dutta
Chief Data Scientist, Spotle.ai
Computer Science, IIT KGP
Rimjhim Ray
Head of Product, Spotle.ai
MBA, SP Jain
Spotle.ai Study Material
Spotle.ai/Learn
Machine Learning is a discipline of Artificial Intelligence. Deep Learning is a
specialised form of Machine Learning.
Artificial
Intelligence
Machine
Learning
Deep
Learning
• Ability of machines to think and act rationallyArtificial
Intelligence
• Ability of machines to take decisions without
being explicitly programmed
• Traditional Machine Learning trains machines
based on labelled and classified data
Machines take decisions basis this data
Machine
Learning
• Machines that mimic human brain through
artificial neural networks capable of extracting
features and classifies from data
• Does not require pre-processing of data or
feature engineering
Deep
Learning
AI, ML and DL – The Overview
Spotle.ai Study Material
Spotle.ai/Learn
" Neural networks are models of the brain
and nervous system
" Highly parallel
" Process information more like the
brain than a serial computer
" Capable of learning – supervised and
unsupervised learning
" Applications:
" As powerful problem solvers
" In simulating complex human
functions like vision, text processing,
complex manoeuvres
Artificial Neural Networks (ANN) - The Machine
Gets A Brain
Deep Learning and
Neural Network
Applications
Spotle.ai Study Material
Spotle.ai/Learn
Confidential
Object Detection – Self-Driving Cars
Spotle.ai Study Material
Spotle.ai/Learn
Confidential
This is a
cat!
This is a
cat!
This is
also a
cat!
Can a computer recognise a cat? Yes ML algorithms can be trained on tagged cat
images, to recognise a cat.
Image Recognition and Classification - Voila! It
Is A Cat!
Spotle.ai Study Material
Spotle.ai/Learn
Confidential
Digital assistant such as Alexa or Siri or the supposedly mind-blowing Google Duplex
use a variety of technology including 1. Speech Processor System consisting of a DSP
or Digital Signal Processing Module 2. Pattern Recognition consisting of NLU (Natural
Language Understanding) and NLP (Natural Language Processing) to understand the
speech and 3. A Speech synthesis module based on NLG or Natural Language
Generation techniques.
Speech Recognition – Google Duplex, Siri,
Alexa
Spotle.ai Study Material
Spotle.ai/Learn 9
" TensorFlow is a deep learning
framework adopted both in research
and production
" Open source software library for
numerical computation using data flow
graphs
" Originally developed by Google Brain
Team to conduct machine learning
research
" “TensorFlow is an interface for
expressing machine learning
algorithms, and an implementation for
executing such algorithms”
What Is TensorFlow?
Spotle.ai Study Material
Spotle.ai/Learn 10
Central Idea: Express a numeric computation as a graph.
● Graph nodes are operations which have any number of inputs
and outputs
● Graph edges are tensors which flow between nodes
TensorFlow – Central Idea
Spotle.ai Study Material
Spotle.ai/Learn 11
" The TensorFlow graph is a description of
computation
" The operators are placed on the graph
nodes
" It executes computations in a given
context of session
" A graph must be launched in a session
for any computation
" A session places the graph operations
onto devices such as CPU and GPU
" A session provides methods to execute
the graph operations
TensorFlow Graphs – Key Points
Spotle.ai Study Material
Spotle.ai/Learn
Estimators
Layers
Python C
Datasets
C++
Metrics
Java
TensorFlow Distributed Engine
High Level APIs
Mid Level APIs
Low Level APIs
TensorFlow Kernel
Estimators are High Level APIs that add an abstraction to graphs and sessions and encapsulate
training, evaluation, prediction simplifying the use of TensorFlow
TensorFlow Architecture
Spotle.ai Study Material
Spotle.ai/Learn
Visualization of graphs which is not an option with Scikit or Numpy
Portability - Computations can be deployed on multiple devices with single API
It is easily trainable for CPUs as well as GPUs for distributed computing
Allows parallel neural network training. Allows batch normalization and optimization of algorithms
Is open-source and backed by a large community of software engineers including Google
13
Why TensorFlow?
Used by:
Spotle.ai Study Material
Spotle.ai/Learn 14
TensorFlow – Example 1
Spotle.ai Study Material
Spotle.ai/Learn 15
TensorFlow – Example 2
Spotle.ai Study Material
Spotle.ai/Learn 16
TensorFlow – Example 2
Spotle.ai Study Material
Spotle.ai/Learn
✓ Understand the concept and architecture of deep learning and neural networks
✓ Learn the TensorFlow and Keras libraries and applications
✓ Learn how to build your own neural networks in TensorFlow and Keras
✓ Understand the various types of neural networks and their applications
✓ Consolidate your career in AI with personalised mentoring from industry
practitioners
✓ Build your own image recognition project using neural networks
✓ Build your career path in advanced AI and Deep Learning
Certificate Masterclass In Advanced Deep
Learning With TensorFlow and Keras
Spotle.ai Study Material
Spotle.ai/Learn
25 Interactive videos with
illustrative examples
24/7 Access to chat and
career support
4
Live classes with
instructors to learn
concepts first-hand
20 Quizzes to test your
learning on the go
9 Hours of lecture
1
Capstone project on
image recognition
using neural networks
Learn Deep Learning With TensorFlow At Spotle: https://spotle.ai/learn/Tensorflow-Keras
Learn At Your Own Pace With Integrated
Instructor Support
Spotle.ai Study Material
Spotle.ai/Learn
19
✓ Certificate of completion
✓ Integrated live projects and performance-based
internships
✓ Membership of Spotle AI Alumni club with priority
access to classes, events and mentoring.
✓ Optimized career path fitment through AI
✓ Personalised career guidance and assistance
Sign up for the Masterclass: https://spotle.ai/learn/Tensorflow-Keras
Your Take - aways

Introduction To TensorFlow

  • 1.
    Spotle.ai Study Material Spotle.ai/Learn ThePower Of Graph Computing For Deep Learning Introduction to
  • 2.
    Spotle.ai Study Material Spotle.ai/Learn2 Instructors Mousum Dutta Chief Data Scientist, Spotle.ai Computer Science, IIT KGP Rimjhim Ray Head of Product, Spotle.ai MBA, SP Jain
  • 3.
    Spotle.ai Study Material Spotle.ai/Learn MachineLearning is a discipline of Artificial Intelligence. Deep Learning is a specialised form of Machine Learning. Artificial Intelligence Machine Learning Deep Learning • Ability of machines to think and act rationallyArtificial Intelligence • Ability of machines to take decisions without being explicitly programmed • Traditional Machine Learning trains machines based on labelled and classified data Machines take decisions basis this data Machine Learning • Machines that mimic human brain through artificial neural networks capable of extracting features and classifies from data • Does not require pre-processing of data or feature engineering Deep Learning AI, ML and DL – The Overview
  • 4.
    Spotle.ai Study Material Spotle.ai/Learn "Neural networks are models of the brain and nervous system " Highly parallel " Process information more like the brain than a serial computer " Capable of learning – supervised and unsupervised learning " Applications: " As powerful problem solvers " In simulating complex human functions like vision, text processing, complex manoeuvres Artificial Neural Networks (ANN) - The Machine Gets A Brain
  • 5.
    Deep Learning and NeuralNetwork Applications
  • 6.
  • 7.
    Spotle.ai Study Material Spotle.ai/Learn Confidential Thisis a cat! This is a cat! This is also a cat! Can a computer recognise a cat? Yes ML algorithms can be trained on tagged cat images, to recognise a cat. Image Recognition and Classification - Voila! It Is A Cat!
  • 8.
    Spotle.ai Study Material Spotle.ai/Learn Confidential Digitalassistant such as Alexa or Siri or the supposedly mind-blowing Google Duplex use a variety of technology including 1. Speech Processor System consisting of a DSP or Digital Signal Processing Module 2. Pattern Recognition consisting of NLU (Natural Language Understanding) and NLP (Natural Language Processing) to understand the speech and 3. A Speech synthesis module based on NLG or Natural Language Generation techniques. Speech Recognition – Google Duplex, Siri, Alexa
  • 9.
    Spotle.ai Study Material Spotle.ai/Learn9 " TensorFlow is a deep learning framework adopted both in research and production " Open source software library for numerical computation using data flow graphs " Originally developed by Google Brain Team to conduct machine learning research " “TensorFlow is an interface for expressing machine learning algorithms, and an implementation for executing such algorithms” What Is TensorFlow?
  • 10.
    Spotle.ai Study Material Spotle.ai/Learn10 Central Idea: Express a numeric computation as a graph. ● Graph nodes are operations which have any number of inputs and outputs ● Graph edges are tensors which flow between nodes TensorFlow – Central Idea
  • 11.
    Spotle.ai Study Material Spotle.ai/Learn11 " The TensorFlow graph is a description of computation " The operators are placed on the graph nodes " It executes computations in a given context of session " A graph must be launched in a session for any computation " A session places the graph operations onto devices such as CPU and GPU " A session provides methods to execute the graph operations TensorFlow Graphs – Key Points
  • 12.
    Spotle.ai Study Material Spotle.ai/Learn Estimators Layers PythonC Datasets C++ Metrics Java TensorFlow Distributed Engine High Level APIs Mid Level APIs Low Level APIs TensorFlow Kernel Estimators are High Level APIs that add an abstraction to graphs and sessions and encapsulate training, evaluation, prediction simplifying the use of TensorFlow TensorFlow Architecture
  • 13.
    Spotle.ai Study Material Spotle.ai/Learn Visualizationof graphs which is not an option with Scikit or Numpy Portability - Computations can be deployed on multiple devices with single API It is easily trainable for CPUs as well as GPUs for distributed computing Allows parallel neural network training. Allows batch normalization and optimization of algorithms Is open-source and backed by a large community of software engineers including Google 13 Why TensorFlow? Used by:
  • 14.
    Spotle.ai Study Material Spotle.ai/Learn14 TensorFlow – Example 1
  • 15.
    Spotle.ai Study Material Spotle.ai/Learn15 TensorFlow – Example 2
  • 16.
    Spotle.ai Study Material Spotle.ai/Learn16 TensorFlow – Example 2
  • 17.
    Spotle.ai Study Material Spotle.ai/Learn ✓Understand the concept and architecture of deep learning and neural networks ✓ Learn the TensorFlow and Keras libraries and applications ✓ Learn how to build your own neural networks in TensorFlow and Keras ✓ Understand the various types of neural networks and their applications ✓ Consolidate your career in AI with personalised mentoring from industry practitioners ✓ Build your own image recognition project using neural networks ✓ Build your career path in advanced AI and Deep Learning Certificate Masterclass In Advanced Deep Learning With TensorFlow and Keras
  • 18.
    Spotle.ai Study Material Spotle.ai/Learn 25Interactive videos with illustrative examples 24/7 Access to chat and career support 4 Live classes with instructors to learn concepts first-hand 20 Quizzes to test your learning on the go 9 Hours of lecture 1 Capstone project on image recognition using neural networks Learn Deep Learning With TensorFlow At Spotle: https://spotle.ai/learn/Tensorflow-Keras Learn At Your Own Pace With Integrated Instructor Support
  • 19.
    Spotle.ai Study Material Spotle.ai/Learn 19 ✓Certificate of completion ✓ Integrated live projects and performance-based internships ✓ Membership of Spotle AI Alumni club with priority access to classes, events and mentoring. ✓ Optimized career path fitment through AI ✓ Personalised career guidance and assistance Sign up for the Masterclass: https://spotle.ai/learn/Tensorflow-Keras Your Take - aways