how machine learning is changing the world
TRANSCRIPT
![Page 1: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/1.jpg)
How Machine Learning is Changing the WorldDEEP LEARNING WITH TENSORFLOW
Emilio Garcia@unindanachado
![Page 3: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/3.jpg)
Agenda
● Motivation● Key concepts
○ AI, ML & DL● Brief revision on ANN
○ Neurons and Layers○ Activation and Loss Functions○ Optimization
● Deep Learning○ Convolutions○ Architectures
● TensorFlow Basis● Demo Time
![Page 4: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/4.jpg)
![Page 5: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/5.jpg)
![Page 6: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/6.jpg)
Key Concepts
Artificial Intelligence
Machine Learning
Deep Learning
“any technique that enables computers to
mimic human intelligence"
“subset of AI that includes abstruse
statistical techniques that enable machines
to improve at tasks with experience"
“algorithms that permit software to train itself
to perform tasks”
![Page 7: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/7.jpg)
Deep Learning = Convolutional Neural Networks
![Page 8: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/8.jpg)
Little History of Neural Networks
1943: McCulloch, W. and Pitts, W. first introduced the idea of a neural network.
1958: Rosenblatt, F introduced backpropagation.
2006: Hinton, G provided a radical new way to train deep neural networks.
Today: Graphic Process Units allow programmers to train networks with several layers.
![Page 9: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/9.jpg)
Float Array
A Typical Neural Network
Input Layer
Hidden Layers (black box)
Output Layer
[input pattern] [output pattern]
many different architectures define the interaction between the input and the
output layer
Float Array
[ -0.025, 0.23, 0.44 ] [ 0.712, 0.471 ]
![Page 10: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/10.jpg)
Considerations
● Input data has to be normalized● Most types of ANNs don’t care about order in train data● Some others do care: BAM (Bidirectional Associative Memory)● Certain Types perform better in certain Problem Domains
![Page 11: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/11.jpg)
Network Types & Problem Domains
Clust Regis Classif Predict Robot Vision Optim
Self-Organizing Map ●●● ● ●
Feedforward ●●● ●●● ●● ●● ●●
Boltzmann Machine ● ●●
Deep Belief Network ●●● ●● ●●
Deep Feedforward ●●● ●●● ●● ●●● ●●
Recurrent Network ●● ●● ●●● ●● ●
Convolutional Network ● ●●● ●●● ●●●
Deep Learning and Neural Networks (Jeff Heaton)
![Page 12: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/12.jpg)
Node, Neuron, Unit
Input 1
Input 2
Input 3
Neuron
Activation Function
Output
Neuron output:
x: inputs
w: weights
: activation function
weight 1 weight 2 weight 3
![Page 13: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/13.jpg)
Neuron Types
I1 I2 B1
B2
B3
N1
N1 N2
O1
N2
Input 1
Hidden1
Hidden2
Context1
Context2
Output1
w2w1
copycopy
w5
w3 w4
w6
Input
Output
BiasHidden
Context
![Page 14: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/14.jpg)
Activation Functions
Linear Threshold
Also called transfer functions. They establish bounds for output of the neurons.
Some of the most popular include:
First used in the original perceptron (McCulloch & Pitts,
1943)
Commonly found in output layers of regression networks
![Page 15: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/15.jpg)
Activation FunctionsSigmoid Hyperbolic Tangent
ReLU
Used to ensure that values are compressed between 0 and 1.
Values range from -1 to 1, mean remains 0. Antisymmetric AFs
yield faster convergence.
Linear, non-saturating function.
![Page 16: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/16.jpg)
● Usually found in the output layer● Represents the probability that the input falls into each class
The Softmax Activation Function
i: index of the output neuron j: indexes of all neurons in the groupz: array of output neurons
![Page 17: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/17.jpg)
Bias
● The weights of the neuron allow us to adjust the slope or shape of the activation function.
● Whereas Bias shift left/right the sigmoid curve.
f(x, 0.5, 0)f(x, 1.0, 0)f(x, 1.5, 0)
f(x, 1.0, 1.0)f(x, 1.0, 0.5)f(x, 1.0, 1.5)
![Page 18: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/18.jpg)
What about convolutions?
“In image processing, a kernel, convolution matrix, or mask is a small matrix. It is useful for blurring, sharpening, embossing, edge
detection, and more. This is accomplished by means of convolution between a kernel and an image.”
-wikipedia-
![Page 19: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/19.jpg)
What about convolutions?
![Page 20: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/20.jpg)
What about convolutions?
![Page 21: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/21.jpg)
What about convolutions?
![Page 22: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/22.jpg)
What about convolutions?
![Page 23: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/23.jpg)
Deep Convolutional Neural Network
![Page 24: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/24.jpg)
Deep Convolutional Neural Network
13 Layer CNN - Alex Krizhevsky (2012)
22 Layer CNN - GoogLeNet: Inception v3 (2014)
![Page 25: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/25.jpg)
Learning to Refine Object Segments - Pedro O. Pinheiro
DeepMask and SharpMask
![Page 26: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/26.jpg)
Demo Time https://github.com/raphsoft/samples/tree/master/meetup/santex-deeplearning
![Page 27: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/27.jpg)
Other Real-World Applications
● Self-Driving Cars● Medical Image Analysis● Bioinformatics● Industry:
○ Churn Prediction○ Sentimental Analysis○ Chatboots○ Recommendation Systems○ Financial Evaluation
● Politics● Security
![Page 28: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/28.jpg)
Questions
![Page 29: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/29.jpg)
Recommended Material & Contact Info
Pattern ClassificationRichard O. Duda
ISBN-13: 978-0471056690ISBN-10: 0471056693
Personal (Work and Academic):[email protected]@pucp.edu.pe
GRPIAA:http://inform.pucp.edu.pe/~grpiaa/ https://www.facebook.com/grpiaa
![Page 30: How machine learning is changing the world](https://reader031.vdocuments.us/reader031/viewer/2022030308/58ed86381a28abef7e8b45db/html5/thumbnails/30.jpg)
Thanks!
We support WarmiLab, join us!
https://www.facebook.com/WarmiLab