(Weird) Introduction to Deep Learning

Favio Vázquez

Data Scientist

@faviovaz

https://github.com/faviovazquez

https://www.linkedin.com/in/faviovazquez/

Webinars

Problem

Top barrier to AI Adoption
Attracting, Acquiring and developing the right AI talent

Right AI talent needs to have following skills:

  1. Deep Learning Concepts
  2. Python Programming
  3. Data libraries like OpenCV, Pandas etc.
  4. DL Frameworks: TensorFlow, Keras, MxNet etc.

Solution

AI Development Tools
(available today)

AI Marketplace
(coming soon)

Deep Learning Studio (DLS)

Deep Learning Concepts:

  1. Python Programming
  2. Data libraries like OpenCV, Pandas etc.
  3. DL Frameworks: TensorFlow, Keras, MxNet etc.
  1. No programming or DL knowledge
  2. Reconfigure and retrain vertical focused AI Apps.
  3. Only needs basic knowledge of data

 

Solution – AI Development Tools

Drag and Drop AI IDE
And Jupyter Notebook
(Available today)

Preconfigured and optimized container environment for different AI frameworks (coming soon)

Outline

What is DL?

What is weird about DL?

Breakthroughs of DL

How to get started with DL?

Doing stuff with DL

Why DL?

Favio Vázquez

About me

  • Venezuelan
  • Physicist and Computer Engineer
  • Master in Physics UNAM
  • Data Scientist
  • Collaborator of Apache Spark project on GitHub and StackOverFlow
  • Sr. Data Scientist Raken Data Group
  • CDS Iron
  • Main Developer of Optimus

Favio Vázquez

  • Very active member of LinkedIn ;)
  • Editor - International Journal of Business Analytics and Intelligence
  • Lecturer Afi (Data Science program-MX)
  • Writer en Towards Data Science, Becoming Human y Planeta Chatbot

About me

Intelligence

Favio Vázquez

Favio Vázquez

Intelligence & AI

What is Intelligence?

Ability to accomplish complex goals.

Lex Fridman - MIT.

Favio Vázquez

We need more than Intelligence.

We need understanding.

We need modeling.

Favio Vázquez

Intelligence & AI

What is Understanding?

Ability to turn complex information into simple, useful information.

Lex Fridman - MIT.

Favio Vázquez

Intelligence & AI

How long for us to be intelligent?

This only took 3.800.000.000 years

Favio Vázquez

And we want this to happen in 5 years...

Favio Vázquez

Intelligence & AI

Kurzweil’s Law of Accelerating Returns

Favio Vázquez

Raymond Kurzweil

Favio Vázquez

Repetitive

Complex

Automation

High

High

Low

Low

Not Common and simple

Not Common and Complex

Common and Simple

Common and Complex

High Level Programming

Conventional Machine Learning​

Deep Learning

Intelligence & AI

Human Intelligence

Favio Vázquez

Intelligence & AI

Achieving Human Intelligence == Singularity of AI

Favio Vázquez

Intelligence & AI

Current state of AI

Favio Vázquez

There will be a breakthrough that will lead us to the singularity

Prophesy of AI

Favio Vázquez

The deep in deep learning isn’t a reference to any kind of deeper understanding achieved by the approach; rather, it stands for this idea of successive layers of representations.

Deep Learning

Favio Vázquez

Learning, in the context of machine learning, describes an automatic search process for better representations.

Deep Learning

Favio Vázquez

Deep Learning

Deep learning is a specific subfield of machine learning, a new take on learning representations from data which puts an emphasis on learning successive “layers” of increasingly meaningful representations.

What is it then??

Favio Vázquez

Artificial Neural Networks

Never say that Deep Learning is just Neural Networks that works like our brain.

Deep Learning

Favio Vázquez

Weird things about Deep Learning

Favio Vázquez

Favio Vázquez

Deep Learning

Normally you have to wait for sometime (sometime years) to be able to digest difficult and advance information in papers or research journals.

DLPaper2Code. arXiv:1711.03543v1

Sethi et al.

Favio Vázquez

Breakthroughs 

of/towards DL

Favio Vázquez

Deep Learning is not all about Neural Networks

Favio Vázquez

Back Propagation

Favio Vázquez

Better initialization of the parameters of the nets.  

This course will teach you the "magic" of getting deep learning to work well. Rather than the deep learning process being a black box, you will understand what drives performance, and be able to more systematically get good results.

Favio Vázquez

Better activation functions. 

This mean, better ways of approximating the functions faster leading to faster training process.

Favio Vázquez

Dropout

(finally not a bad thing to say you did).

Better ways of preventing overfitting and more.

Favio Vázquez

Convolutional Neural Nets

Favio Vázquez

Residual Nets

Favio Vázquez

Region Based CNNs

Favio Vázquez

Recurrent Neural Networks (RNNs) and LSTMs.

DeepBach.  arXiv:1612.01010v2.

Hadjeres, Pachet and Nielsen

Favio Vázquez

BTW: It was shown by Liao and Poggio (2016) that

ResNet == RNN

 arXiv:1604.03640v1

Favio Vázquez

Generative Adversarial Networks (GANs)

Favio Vázquez

Capsule Networks (CapsNet)

Favio Vázquez

WAIT!

What about Big Data and Data Science?

Favio Vázquez

Big Data + Data Science + AI current state

Favio Vázquez

Big Data is one of the Catalysers to get to Artificial General Intelligence

What else do we need?

Favio Vázquez

 It enable us to create intelligence through AI

Favio Vázquez

Data Science!

Getting Started with Deep Learning

Favio Vázquez

Favio Vázquez

Favio Vázquez

Favio Vázquez

Siraj Raval

Favio Vázquez

Brandon Rohrer

Favio Vázquez

Lex Fridman

Favio Vázquez

François Chollet

Favio Vázquez

IBM

Favio Vázquez

Favio Vázquez

Favio Vázquez

Favio Vázquez

Doing stuff with DL

Favio Vázquez

Programming (Learn!!)

Favio Vázquez

AutoML+GUI+Notebook

Favio Vázquez

DEMO

Questions?

Favio Vázquez

Data Scientist

@faviovaz

Made with Slides.com