Understanding Artificial

Neural Networks
 ANN

● A large part of this section will focus on

theory behind many of the ideas we will
implement with code.
● Let’s do a quick review of how we will
gradually build an understanding of
artificial neural networks.
 ANN

● Theory Topics
○ Perceptron Model to Neural Networks
○ Activation Functions
○ Cost Functions
○ Feed Forward Networks
○ BackPropagation
 ANN

● Coding Topics
○ TensorFlow 2.0 Keras Syntax
○ ANN with Keras
■ Regression
■ Classification
○ Exercises for Keras ANN
○ Tensorboard Visualizations
Let’s get started!
Perceptron Model
 Perceptron model

● To begin understanding deep learning, we

will build up our model abstractions:
○ Single Biological Neuron
○ Perceptron
○ Multi-layer Perceptron Model
○ Deep Learning Neural Network
 Perceptron model

● As we learn about more complex models,

we’ll also introduce concepts, such as:
○ Activation Functions
○ Gradient Descent
○ BackPropagation
 Perceptron model

● If the whole idea behind deep learning is to

have computers artificially mimic biological
natural intelligence, we should probably
build a general understanding of how
biological neurons work!
 Perceptron model

● Stained Neurons in cerebral cortex

 Perceptron model
● Illustration of biological neurons
 Perceptron model
● Illustration of biological neurons
 Perceptron model
● Let’s really simplify this!
 Perceptron model
● Simplified Biological Neuron Model

Dendrites
Axon

Nucleus
 Perceptron model

● A perceptron was a form of neural network

introduced in 1958 by Frank Rosenblatt.
● Amazingly, even back then he saw huge
potential:
○ "...perceptron may eventually be able
to learn, make decisions, and translate
languages."
 Perceptron model

● However, in 1969 Marvin Minsky and

Seymour Papert's published their book
Perceptrons.
● It suggested that there were severe
limitations to what perceptrons could do.
● This marked the beginning of what is
known as the AI Winter, with little funding
into AI and Neural Networks in the 1970s.
 Perceptron model

● Fortunately for us, we now know the

amazing power of neural networks, which
all stem from the simple perceptron model,
so let’s head back and convert our simple
biological neuron model into the
perceptron model.
 Perceptron model
● Perceptron Model

Dendrites
Axon

Nucleus
 Perceptron model
● Perceptron Model

Inputs Output
 Perceptron model
● Let’s work through a simple example

x1
Inputs Output

x2
 Perceptron model
● Let’s work through a simple example

x1
Inputs f(X) Output

x2
 Perceptron model
● Let’s work through a simple example

x1
y
Inputs f(X) Output

x2
 Perceptron model
● If f(X) is just a sum, then y=x1+x2

x1
y
Inputs f(X) Output

x2
 Perceptron model
● Realistically, we would want to be able to
adjust some parameter in order to “learn”
x1
y
Inputs f(X) Output

x2
 Perceptron model
● Let’s add an adjustable weight we multiply
against x
w1
x1
y
Inputs f(X) Output
w2

x2
 Perceptron model
● Now y = x1w1 + x2w2

w1
x1
y
Inputs f(X) Output
w2

x2
 Perceptron model
● We could update the weights to effect y

w1
x1
y
Inputs f(X) Output
w2

x2
 Perceptron model
● But what if an x is zero? w won’t change
anything!
w1
x1
y
Inputs f(X) Output
w2

x2
 Perceptron model
● Let’s add in a bias term b to the inputs.

w1
x1
y
Inputs f(X) Output
w2

x2
 Perceptron model
● Let’s add in a bias term b to the inputs.

*w1 + b
x1
y
Inputs f(X) Output

x2 *w2 + b
 Perceptron model
● y = (x1w1 + b) + (x2w2 + b)

*w1 + b
x1
y
Inputs f(X) Output

x2 *w2 + b
 Perceptron model
● We can expand this to a generalization:

x1
*w1 + b y
Inputs f(X) Output
x2 *w2 + b

xn *wn + b
 Perceptron model

● We’ve been able to model a biological

neuron as a simple perceptron!
Mathematically our generalization was:
 Perceptron model

● Later on we will see how we can expand

this model to have X be a tensor of
information ( an n-dimensional matrix).
 Perceptron model

● Let’s review what we learned:

○ We understand the very basics of a
biological neuron
○ We saw how we can create a simple
perceptron model replicating the core
concepts behind a neuron.
Neural Networks
 Neural Networks

● A single perceptron won’t be enough to

learn complicated systems.
● Fortunately, we can expand on the idea of a
single perceptron, to create a multi-layer
perceptron model.
 Neural Networks

● A single perceptron won’t be enough to

learn complicated systems.
● Fortunately, we can expand on the idea of a
single perceptron, to create a multi-layer
perceptron model.
● We’ll also introduce the idea of activation
functions.
 Neural Networks

● To build a network of perceptrons, we can

connect layers of perceptrons, using a
multi-layer perceptron model.
 Neural Networks

● The outputs of one perceptron are directly

fed into as inputs to another perceptron.
 Neural Networks

● This allows the network as a whole to learn

about interactions and relationships
between features.
 Neural Networks

● The first layer is the input layer

 Neural Networks

● The last layer is the output layer.

● Note: This last layer can be more than one
neuron
 Neural Networks

● Layers in between the input and output

layers are the hidden layers.
 Neural Networks

● Hidden layers are difficult to interpret, due

to their high interconnectivity and distance
away from known input or output values.
 Neural Networks

● Neural Networks become “deep neural

networks” if then contain 2 or more
hidden layers.
 Neural Networks

● Neural Networks become “deep neural

networks” if then contain 2 or more
hidden layers.
 Neural Networks

● Terminology:
○ Input Layer: First layer that directly
accepts real data values
○ Hidden Layer: Any layer between input
and output layers
○ Output Layer: The final estimate of the
output.
 Neural Networks

● What is incredible about the neural

network framework is that it can be used
to approximate any function.
● Zhou Lu and later on Boris Hanin proved
mathematically that Neural Networks can
approximate any convex continuous
function.
 Neural Networks

● For more details on this check out the

Wikipedia page for “Universal
Approximation Theorem”
 Neural Networks

● Previously in our simple model we saw that

the perceptron itself contained a very
simple summation function f(x).
● For most use cases however that won’t be
useful, we’ll want to be able to set
constraints to our output values, especially
in classification tasks.
 Neural Networks

● In a classification tasks, it would be useful

to have all outputs fall between 0 and 1.
● These values can then present probability
assignments for each class.
● In the next lecture, we’ll explore how to use
activation functions to set boundaries to
output values from the neuron.
Activation Functions
 Neural Networks

● Recall that inputs x have a weight w and a

bias term b attached to them in the
perceptron model.
● Which means we have
○ x*w + b
 Neural Networks

● Which means we have

○ x*w + b
● Clearly w implies how much weight or
strength to give the incoming input.
● We can think of b as an offset value,
making x*w have to reach a certain
threshold before having an effect.
 Neural Networks

● For example if b= -10

○ x*w + b
● Then the effects of x*w won’t really start
to overcome the bias until their product
surpasses 10.
● After that, then the effect is solely based
on the value of w.
● Thus the term “bias”
 Neural Networks

● Next we want to set boundaries for the

overall output value of:
○ x*w + b
● We can state:
○ z = x*w + b
● And then pass z through some
activation function to limit its value.
 Neural Networks

● A lot of research has been done into

activation functions and their effectiveness.
● Let’s explore some common activation
functions.
 Perceptron model
● Recall our simple perceptron has an f(X)

w1
x1
y
Inputs f(X) Output
w2
+b
x2
 Perceptron model
● If we had a binary classification problem,
we would want an output of either 0 or 1.
w1
x1
y
Inputs f(X) Output
w2
+b
x2
 Neural Networks

● To avoid confusion, let’s define the total

inputs as a variable z.
● Where z = wx + b
● In this context, we’ll then refer to activation
functions as f(z).
● Keep in mind, you will often see these
variables capitalized f(Z) or X to denote a
tensor input consisting of multiple values.
 Deep Learning

● The most simple networks rely on a basic

step function that outputs 0 or 1.
1
Output

0
0
z = wx + b
 Deep Learning

● Regardless of the values, this always

outputs 0 or 1.
1
Output

0
0
z = wx + b
 Deep Learning

● This sort of function could be useful for

classification (0 or 1 class).
1
Output

0
0
z = wx + b
 Deep Learning

● However this is a very “strong” function,

since small changes aren’t reflected.
1
Output

0
0
z = wx + b
 Deep Learning

● There is just an immediate cut off that

splits between 0 and 1.
1
Output

0
0
z = wx + b
 Deep Learning

● It would be nice if we could have a more

dynamic function, for example the red line!
1
Output

0
0
z = wx + b
 Deep Learning

● Lucky for us, this is the sigmoid function!

1
Output

0
0
z = wx + b
 Deep Learning

● Changing the activation function used can

be beneficial depending on the task!
1
Output

0
0
z = wx + b
 Deep Learning

● This still works for classification, and will be

more sensitive to small changes.
1
Output

0
0
z = wx + b
 Deep Learning

● Let’s discuss a few more activation

functions that we’ll encounter!
1
Output

0
0
z = wx + b
 Deep Learning

● Hyperbolic Tangent: tanh(z)

1
Output

-1
0
 Deep Learning

● Hyperbolic Tangent: tanh(z)

● Outputs between -1 and 1 instead of 0 to 1
1
Output

-1
0
 Deep Learning

● Rectified Linear Unit (ReLU): This is

actually a relatively simple function:
max(0,z)
Output

0
z = wx + b
 Deep Learning

● ReLu has been found to have very good

performance, especially when dealing with
the issue of vanishing gradient.
● We’ll often default to ReLu due to its
overall good performance.
 Deep Learning

● For a full list of possible activation functions

check out:
● en.wikipedia.org/wiki/Activation_function
 Multi-Class
Activation Functions
 Deep Learning

● Notice all these activation functions make

sense for a single output, either a
continuous label or trying to predict a
binary classification (either a 0 or 1).
● But what should we do if we have a
multi-class situation?
 Deep Learning

● There are 2 main types of multi-class

situations
○ Non-Exclusive Classes
■ A data point can have multiple
classes/categories assigned to it
○ Mutually Exclusive Classes
■ Only one class per data point.
 Deep Learning

● Non-Exclusive Classes
■ A data point can have multiple
classes/categories assigned to it
○ Photos can have multiple tags (e.g.
beach, family, vacation, etc…)
 Deep Learning

● Mutually Exclusive Classes

■ A data point can only have one
class/category assigned to it
○ Photos can be categorized as being in
grayscale (black and white) or full color
photos. A photo can not be both at the
same time.
 Deep Learning

● Organizing Multiple Classes

○ The easiest way to organize multiple
classes is to simply have 1 output node
per class.
 Neural Networks

● Previously we thought of the last output

layer as a single node.
 Neural Networks

● This single node could output a continuous

regression value or binary classification (0
or 1).
 Neural Networks

● Let’s expand this output layer to work for

the case of multi-classification.
 Multiclass Classification

● Organizing for Multiple Classes

Hidden Layers
 Multiclass Classification

● Organizing for Multiple Classes

Class One

Class Two

Hidden Layers
Class N
 Deep Learning

● Organizing Multiple Classes

○ This means we will need to organize
categories for this output layer.
○ We can’t just have categories like “red”,
“blue”, “green”, etc...
 Deep Learning

● Organizing Multiple Classes

○ Instead we use one-hot encoding
○ Let’s take a look at what this looks like
for mutually exclusive classes.
 Deep Learning

● Mutually Exclusive Classes

Data Point 1 RED

Data Point 2 GREEN

Data Point 3 BLUE

... ...

Data Point N RED

 Deep Learning

● Mutually Exclusive Classes

RED GREEN BLUE
Data Point 1 RED
Data Point 1 1 0 0
Data Point 2 GREEN
Data Point 2 0 1 0
Data Point 3 BLUE
Data Point 3 0 0 1
... ...
... ... ... ...
Data Point N RED
Data Point N 1 0 0
 Deep Learning

● Non-Exclusive Classes
A B C
Data Point 1 A,B
Data Point 1 1 1 0
Data Point 2 A
Data Point 2 1 0 0
Data Point 3 C,B
Data Point 3 0 1 1
... ...
... ... ... ...
Data Point N B
Data Point N 0 1 0
 Deep Learning

● Now that we have our data correctly

organized, we just need to choose the
correct classification activation function
that the last output layer should have.
 Deep Learning

● Non-exclusive
○ Sigmoid function
■ Each neuron will output a value
between 0 and 1, indicating the
probability of having that class
assigned to it.
 Multiclass Classification

● Sigmoid Function for Non-Exclusive Classes

Class One

Class Two

Hidden Layers
Class N
 Multiclass Classification

● Sigmoid Function for Non-Exclusive Classes

1
Class One 0.8
0

1
Class Two 0.2
0

Hidden Layers
1
Class N 0.3
0
 Multiclass Classification

● Sigmoid Function for Non-Exclusive Classes

1
Class One 0.8
0

1
Class Two 0.6
0

Hidden Layers
1
Class N 0.2
0
 Deep Learning

● Non-exclusive
○ Sigmoid function
■ Keep in mind this allows each neuron
to output independent of the other
classes, allowing for a single data
point fed into the function to have
multiple classes assigned to it.
 Deep Learning

● Mutually Exclusive Classes

○ But what do we do when each data
point can only have a single class
assigned to it?
○ We can use the softmax function for
this!
 Deep Learning

● Softmax Function
 Deep Learning

● Mutually Exclusive Classes

○ Softmax function calculates the
probabilities distribution of the event
over K different events.
○ This function will calculate the
probabilities of each target class over all
possible target classes.
 Deep Learning

● Mutually Exclusive Classes

○ The range will be 0 to 1, and the sum of
all the probabilities will be equal to
one.
○ The model returns the probabilities of
each class and the target class chosen
will have the highest probability.
 Deep Learning

● Mutually Exclusive Classes

○ The main thing to keep in mind is that if
you use softmax for multi-class
problems you get this sort of output:
■ [Red , Green , Blue]
■ [ 0.1 , 0.6 , 0.3 ]
 Deep Learning

● Mutually Exclusive Classes

○ The probabilities for each class all sum
up to 1. We choose the highest
probability as our assignment.
■ [Red , Green , Blue]
■ [ 0.1 , 0.6 , 0.3 ]
 Deep Learning

● Review
○ Perceptrons expanded to neural
network model
○ Weights and Biases
○ Activation Functions
○ Time to learn about Cost Functions!
Cost Functions and
Gradient Descent
 Deep Learning

● We now understand that neural

networks take in inputs, multiply them
by weights, and add biases to them.
● Then this result is passed through an
activation function which at the end of
all the layers leads to some output.
 Deep Learning

● This output ŷ is the model’s estimation of

what it predicts the label to be.
● So after the network creates its
prediction, how do we evaluate it?
● And after the evaluation how can we
update the network’s weights and
biases?
 Deep Learning

● We need to take the estimated outputs

of the network and then compare them
to the real values of the label.
● Keep in mind this is using the training
data set during the fitting/training of the
model.
 Deep Learning

● The cost function (often referred to as a

loss function) must be an average so it
can output a single value.
● We can keep track of our loss/cost during
training to monitor network
performance.
 Deep Learning

● We’ll use the following variables:

○ y to represent the true value
○ a to represent neuron’s prediction
● In terms of weights and bias:
○ w*x + b = z
○ Pass z into activation function σ(z) = a
 Deep Learning

● One very common cost function is the

quadratic cost function:
 Deep Learning

● We simply calculate the difference

between the real values y(x) against our
predicted values a(x).
 Deep Learning

● Note: The notation shown here

corresponds to vector inputs and
outputs, since we will be dealing with a
batch of training points and predictions.
 Deep Learning

● Notice how squaring this does 2 useful

things for us, keeps everything positive
and punishes large errors!
 Deep Learning

● We can think of the cost function as:

 Deep Learning

● W is our neural network's weights, B is

our neural network's biases, Sr is the
input of a single training sample, and Er
is the desired output of that training
sample.
 Deep Learning

● Notice how that information was all

encoded in our simplified notation.
● The a(x) holds information about
weights and biases.
 Deep Learning

● This means that if we have a huge

network, we can expect C to be quite
complex, with huge vectors of weights
and biases.
 Deep Learning

● Here is a small network with all its

parameters labeled:
 Deep Learning

● That is a lot to calculate!

● How do we solve this?
 Deep Learning

● In a real case, this means we have some

cost function C dependent lots of
weights!
○ C(w1,w2,w3,....wn)
● How do we figure out which weights
lead us to the lowest cost?
 Deep Learning

● For simplicity, let’s imagine we only had

one weight in our cost function w.
● We want to minimize our loss/cost
(overall error).
● Which means we need to figure out
what value of w results in the minimum
of C(w)
 Deep Learning

● Our “simple” function C(w)

C(w)

w
 Deep Learning

● What value of w minimizes our cost?

C(w)

w
 Deep Learning

● What value of w minimizes our cost?

C(w)

w
 Deep Learning

● What value of w minimizes our cost?

C(w)

wmin
 Deep Learning

● Students of calculus know we could take

a derivative and solve for 0.
C(w)

wmin
 Deep Learning

● But recall our real cost function will be

very complex!
C(w)

w
 Deep Learning

● And it will be n-dimensional!

C(w)

w
 Deep Learning

● And it will be n-dimensional since our

networks will have 1000s of weights
C(w)

w
 Deep Learning

● We can use gradient descent to solve

this problem.
C(w)

w
 Deep Learning

● Let’s go back to our simplified version to

see how this works.
C(w)

wmin
 Deep Learning

● We can calculate the slope at a point

C(w)

wmin
 Deep Learning

● We can calculate the slope at a point

C(w)

wmin
 Deep Learning

● Then we move in the downward

direction of the slope.
C(w)

wmin
 Deep Learning

● Then we move in the downward

direction of the slope.
C(w)

wmin
 Deep Learning

● Then we move in the downward

direction of the slope.
C(w)

wmin
 Deep Learning

● Until we converge to zero, indicating a

minimum.
C(w)

wmin
 Deep Learning

● We could have changed our step size to

find the next point!
C(w)

wmin
 Deep Learning

● Our steps:
C(w)

wmin
 Deep Learning

● Smaller steps sizes take longer to find

the minimum.
C(w)

wmin
 Deep Learning

● Larger steps are faster, but we risk

overshooting the minimum!
C(w)

wmin
 Deep Learning

● This step size is known as the learning

rate.
C(w)

wmin
 Deep Learning

● The learning rate we showed in our

illustrations was constant (each step size
was equal)
● But we can be clever and adapt our step
size as we go along.
 Deep Learning

● We could start with larger steps, then go

smaller as we realize the slope gets
closer to zero.
● This is known as adaptive gradient
descent.
 Deep Learning

● In 2015, Kingma and Ba published their

paper: “Adam: A Method for Stochastic
Optimization“.
● Adam is a much more efficient way of
searching for these minimums, so you
will see us use it for our code!
 Deep Learning

● Adam versus other gradient descent

algorithms:
 Deep Learning

● Realistically we’re calculating this

descent in an n-dimensional space for all
our weights.
 Deep Learning

● When dealing with these N-dimensional

vectors (tensors), the notation changes
from derivative to gradient.
● This means we calculate
∇C(w1,w2,...wn)
 Deep Learning

● For classification problems, we often use

the cross entropy loss function.
● The assumption is that your model
predicts a probability distribution p(y=i)
for each class i=1,2,…,C.
 Deep Learning

● For a binary classification this results in:

● For M number of classes > 2

 Deep Learning

● Review:
○ Cost Functions
○ Gradient Descent
○ Adam Optimizer
○ Quadratic Cost and Cross-Entropy
 Deep Learning

● So far we understand how networks can

take in input , effect that input with
weights, biases, and activation functions
to produce an estimated output.
● Then we learned how to evaluate that
output.
 Deep Learning

● The last thing we need to learn about

theory is:
○ Once we get our cost/loss value,
how do we actually go back and
adjust our weights and biases?
● This is backpropagation, and it is what
we are going to cover next!
Backpropagation
 Deep Learning

● The last theory topic we will cover is

backpropagation.
● We’ll start by building an intuition
behind backpropagation, and then we’ll
dive into the calculus and notation of
backpropagation.
 Deep Learning

● Fundamentally, we want to know how

the cost function results changes with
respect to the weights in the network, so
we can update the weights to minimize
the cost function
 Deep Learning

● Let’s begin with a very simple network,

where each layer only has 1 neuron
 Deep Learning

● Each input will receive a weight and bias

w1 +b1 w2 +b2 w3 +b3

 Deep Learning

● This means we have:

○ C(w1,b1,w2,b2,w3,b3)

w1 +b1 w2 +b2 w3 +b3

 Deep Learning

● We’ve already seen how this process

propagates forward.
● Let’s start at the end to see the
backpropagation.
w1 +b1 w2 +b2 w3 +b3
 Deep Learning

● Let’s say we have L layers, then our

notation becomes:

L-n L-2 L-1 L

 Deep Learning

● Focusing on these last two layers, let’s

define z=wx+b
● Then applying an activation function
we’ll state: a = σ(z)

L-1 L
 Deep Learning

● This means we have:

○ zL = wL aL-1 +bL

L-1 L
 Deep Learning

● This means we have:

○ zL = wL aL-1 +bL
○ aL = σ(zL)

L-1 L
 Deep Learning

● This means we have:

○ zL = wL aL-1 +bL
○ aL = σ(zL)
○ C0(...) =(aL - y)2

L-1 L
 Deep Learning

● We want to understand how sensitive is

the cost function to changes in w:

L-1 L
 Deep Learning

● Using the relationships we already know

along with the chain rule:

L-1 L
 Deep Learning

● We can calculate the same for the bias

terms:

L-1 L
 Deep Learning

● The main idea here is that we can use

the gradient to go back through the
network and adjust our weights and
biases to minimize the output of the
error vector on the last output layer.
 Deep Learning

● Using some calculus notation, we can

expand this idea to networks with
multiple neurons per layer.
● Hadamard Product
 Deep Learning

● Given this notation and

backpropagation, we have a few main
steps to training neural networks.
● Note! You do not need to fully
understand these intricate details to
continue with the coding portions.
 Deep Learning

● Step 1: Using input x set the activation

function a for the input layer.
○ z = w x +b
○ a = σ(z)
● This resulting a then feeds into the next
layer (and so on).
 Deep Learning

● Step 2: For each layer, compute:

○ zL = wL aL-1 +bL
○ aL = σ(zL)
 Deep Learning

● Step 3: We compute our error vector:

○ δL=∇aC⊙σ′(zL)
 Deep Learning

● Step 3: We compute our error vector:

○ δL=∇aC⊙σ′(zL)
■ ∇aC=(aL−y)
■ Expressing the rate of change of C
with respect to the output
activations
 Deep Learning

● Step 3: We compute our error vector:

○ δL=(aL−y)⊙σ′(zL)
 Deep Learning

● Step 3: We compute our error vector:

○ δL=(aL−y)⊙σ′(zL)
● Now let’s write out our error term for a
layer in terms of the error of the next
layer (since we’re moving backwards).
● Font Note: lowercase L
● Font Note: Number 1
 Deep Learning

● Step 4: Backpropagate the error:

○ For each layer: L−1,L−2,… we compute
(note the lowercase L (l)):
■ δl=(wl+1)Tδl+1⊙σ′(zl)
■ (wl+1)T is the transpose of the
weight matrix of l+1 layer
 Deep Learning

● Step 4: Backpropagate the error:

○ This is the generalized error for any
layer l:
■ δl=(wl+1)Tδl+1⊙σ′(zl)
■ (wl+1)T is the transpose of the
weight matrix of L+1 layer
 Deep Learning

● Step 4: When we apply the transpose

weight matrix, (wl+1)T we can think
intuitively of this as moving the error
backward through the network, giving
us some sort of measure of the error at
the output of the lth layer.
 Deep Learning

● Step 4: We then take the Hadamard

product ⊙σ′(zl). This moves the error
backward through the activation
function in layer l, giving us the error δl in
the weighted input to layer l.
 Deep Learning

● The gradient of the cost function is given

by:
○ For each layer: L−1,L−2,… we compute
 Deep Learning

● This then allows us to adjust the weights

and biases to help minimize that cost
function.
● Check out the external links for more
details!
TensorFlow and Keras
 Deep Learning

● Before we begin learning how to code

our own neural networks, let’s quickly
clarify the differences between
TensorFlow and Keras!
 Deep Learning

● TensorFlow is an open-source deep

learning library developed by Google,
with TF 2.0 being officially released in
late 2019.
 Deep Learning

● TensorFlow has a large ecosystem of

related components, including libraries
like Tensorboard, Deployment and
Production APIs, and support for various
programming languages.
 Deep Learning

● Keras is a high-level python library that

can use a variety of deep learning
libraries underneath, such as:
TensorFlow, CNTK, or Theano.
 Deep Learning

● TensorFlow 1.x had a complex python

class system for building models, and
due to the huge popularity of Keras,
when TF 2.0 was released, TF adopted
Keras as the official API for TF.
 Deep Learning

● While Keras still also remains as a

separate library from Tensorflow, it can
also now officially be imported through
TF, so there is now need to additionally
install it.
 Deep Learning

● The Keras API is easy to use and builds

models by simply adding layers on top of
each other through simple calls.
● Let’s now explore the basics of the Keras
API for TensorFlow!
 Keras Classification
Code Along - Part One
 Deep Learning

● This lecture will show how to perform a

classification task with TensorFlow.
● We will also focus on how to identify and
deal with overfitting through Early
Stopping Callbacks and Dropout Layers.
 Deep Learning

● Early Stopping
○ Keras can automatically stop training
based on a loss condition on the
validation data passed during the
model.fit() call.
 Deep Learning

● Dropout Layers
○ Dropout can be added to layers to
“turn off” neurons during training to
prevent overfitting.
 Deep Learning

● Dropout Layers
○ Each Dropout layer will “drop” a
user-defined percentage of neuron
units in the previous layer every batch.
Introduction to Project
 Deep Learning

● It’s time for a project! You will build a

model that will attempt to predict
whether or not someone will pay back
their loan based on historical
information.
 Deep Learning

● You have 3 options for this project

○ Completely Solo
○ Exercise Guide Notebook
○ Code Along with Solution Lectures
 Deep Learning

● Completely Solo
○ Read the introduction in the Exercise
notebook, then proceed with your own
methods to build a predictive model.
 Deep Learning

● Exercise Guide Notebook

○ Follow the written steps in the
notebook to complete tasks that guide
you through building a predictive
model.
 Deep Learning

● Code along with solutions lecture

○ Skip the next overview lecture and
code along with us as we guide you
through our solution video for the
project.
 Deep Learning

● This is a large project!

● To reflect a realistic situation, we will
spend a lot of time performing feature
engineering and analyzing our data.
● Let’s explore the project guide notebook
in the next lecture!
 Keras Project
Exercise Solutions
Exploratory Data Analysis
 Keras Project
Exercise Solutions
Data Preprocessing: Missing Data
 Keras Project
Exercise Solutions
Categorical Data
 Keras Project
Exercise Solutions
Data Preprocessing
 Keras Project
Exercise Solutions
Creating and Training a Model
 Keras Project
Exercise Solutions
Model Evaluation
 Keras Project
Exercise Solutions
Model Evaluation
Tensorboard
 Deep Learning

● Tensorboard is a visualization tool from

Google designed to work in conjunction
with TensorFlow to visualize various
aspects of your model.
 Deep Learning

● Here we will simply understand how to

view the Tensorboard dashboard in our
browser and analyze an existing model.
● NOTE - This lecture requires that you
understand file paths and the location
of your notebook or .py file!
 Deep Learning

● Keep in mind, Tensorboard is a separate

library from TensorFlow.
● Google Collab Users can follow with
Google’s official guide and pre-made
notebook:
https://www.tensorflow.org/tensorboard/tensorboard_in_notebooks