EE5075
Artificial Intelligence Applications
LECTURE 4
Machine Learning:
Implementation of Regression and
Classification using ANN
By Buddhika Jayasekara ([email protected])
Recap - Artificial Neuron
✓ Neural computing requires a
number of neurons, to be
connected together into a Inputs Weights
neural network. p1 w1
✓ Each neuron within the w2
a
network is usually a simple
p2
w3 f Output
processing unit which takes p3
one or more inputs and 1
produces an output. Bias
a = f (p1 w1 + p2 w2 + p3 w3 + b ) = f ( pi wi + b )
✓ At each neuron, every input
has an associated weight
which modifies the strength of
each input.
2
Recap - Artificial Neuron
3
Recap - Artificial Neuron
4
Recap - Artificial Neuron
5
Recap - Artificial Neuron
6
Recap - Artificial Neuron
7
Recap - Artificial Neuron
8
Recap - Feed-forward Neural Networks
• Information flow is unidirectional
• Data is presented to Input layer
• Passed on to Hidden Layer
• Passed on to Output layer
• Information processing is parallel
Internal representation
(interpretation) of data
9
Recap - Feedforward neural network
✓ Use data set [x,t] = simplefit_dataset;
✓ Plot(x,t)
10
8
Tartget values
X 6 t
Input Target
space 4
space
2
0
0 1 2 3 4 5 6 7 8 9 10
X values
Neural Network
Feed forward neural network
10
Recap - Artificial Neural Network: Learning
11
Recap - Artificial Neural Network: Learning
12
Recap - Artificial Neural Network: Learning
13
Recap - Gradient Descent Algorithm
14
Machine Learning
8/31/2024 7:26 AM 15
Machine Learning - Applications
8/31/2024 7:26 AM 16
What is Deep Learning?
8/31/2024 7:26 AM 17
Activation Functions
8/31/2024 7:26 AM 18
Activation Functions
8/31/2024 7:26 AM 19
Activation Functions
8/31/2024 7:26 AM 20
Activation Functions
8/31/2024 7:26 AM 21
Activation Functions
8/31/2024 7:26 AM 22
Activation Function - Selection
8/31/2024 7:26 AM 23
Recap - Getting Familiar with Libraries
24
Deep Learning Frameworks
8/31/2024 7:26 AM 25
TensorFlow Keras Module
8/31/2024 7:26 AM 26
8/31/2024 7:26 AM 27
8/31/2024 7:26 AM 28
Machine Learning - Regression
✓ Fitting line or curve to set of data points
✓ Supervised learning
29
Machine Learning -Classification
✓ Predict a category or label
✓ Regression –predict a
number
✓ Supervised learning
✓ E.g line to separate data
points
30
Machine Learning -Classification
✓ Decision boundary may not be linear,
▪ line or hyperplane
31
Classification models
✓ Use of Line for 2D Classification
𝑌 = 𝑚𝑋 + 𝑏
32
Classification
33
Classification
34
8/31/2024 7:26 AM 35
8/31/2024 7:26 AM 36
8/31/2024 7:26 AM 37
8/31/2024 7:26 AM 38
8/31/2024 7:26 AM 39
Student Activity
✓ Download the workbook and follow the class activities
8/31/2024 7:26 AM 40
8/31/2024 7:26 AM 41
8/31/2024 7:26 AM 42
8/31/2024 7:26 AM 43
8/31/2024 7:26 AM 44
8/31/2024 7:26 AM 45
8/31/2024 7:26 AM 46
8/31/2024 7:26 AM 47
8/31/2024 7:26 AM 48
8/31/2024 7:26 AM 49
8/31/2024 7:26 AM 50
8/31/2024 7:26 AM 51
8/31/2024 7:26 AM 52
8/31/2024 7:26 AM 53
Example -Regression
54
55
56
57
Self Study and Assignment Q1 - Classification
Example
✓ Use sklearn.datasets
load_breast_cancer
Ref: https://scikit-learn.org/stable/datasets/toy_dataset.html
58
59
Benign: These are not cancerous.
Malignant: These are cancerous.
60
NN Model
61
62
https://scikit-
learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html
63
Additional References
✓ tf.keras.layers.Dense:
https://www.tensorflow.org/api_docs/python/tf/keras/layers/De
nse
✓ tf.keras.optimizers.Adam:
https://www.tensorflow.org/api_docs/python/tf/keras/optimizer
s/Adam
✓ tf.keras.losses.BinaryCrossentropy:
https://www.tensorflow.org/api_docs/python/tf/keras/losses/Bi
naryCrossentropy
64
The Sequential model - tf.keras.Sequential
✓ Ref: https://www.tensorflow.org/guide/keras/sequential_model
65
NN Model and Training
A Neuron
66
NN Model and Training
67
68
69
Use the NN model for Prediction
70
71
72
Assignment Q1 - Classification
✓ Use sklearn.datasets
And select any other dataset and develop an
ANN with hidden layer with many neurons
Ref: https://scikit-learn.org/stable/datasets/toy_dataset.html
73
Self Study
✓ Logistic regression vs Linear regression
✓ Use of logistic regression for classification
✓ Loss functions
▪ Binary cross entropy
▪ Mean squared error
74
Thank You
8/31/2024 7:26 AM 75
