CAI1001C

AI Thinking
Elements of ML
Disclaimer

Miami Dade College has adapted the content on the subsequent slides from the
AI for Current Workforce program, which is part of
Intel® Digital Readiness Programs.
Topics

1. Introduction to Machine Learning.

2. What are the different Machine Learning algorithms and their applications?
3. What is the difference between Machine Learning and Deep Learning?
Let’s start by understanding what is
Machine Learning.

4
Introduction to Machine Learning
Introduction to Machine Learning

• What is Machine Learning?

• How Machine Learning differs from Traditional Programming?

• History of Machine Learning

• How Machine Learning works?

• Why Machine Learning?

What is
Machine
Learning?
What is Machine Learning?

“Field of study that gives computers the ability to

learn without being explicitly programmed”
What is Machine Learning?

Example

Given the history of home sales in a city, you could use machine learning to
create a model that is able to predict the value of a different home in that
same city.
How ML differs from Traditional Programming?

Data + Program Traditional Programming Output

Machine Learning Model

Data (input and output)
How ML differs from Traditional Programming?
 Let us now look at the history of
Machine Learning and what it has
achieved so far.

12
 History of Machine Learning [2]
Alexey
Ivakhnen
ko and Tin Kam
Arthur Ho Fei-Fei Li
Valentin created
First Samuel publishe
Lapa ImageNet, a
mathemat created d paper
created database of
ical model Checker on
first labelled
of Neural s random
multi- images
Network Progra decision
layer
m perceptro forests
n
199 201
1950 1959 1967 7 6

1943 1956 1965 199 200

5 9

Thomas IBM AlphaGo

Alan Turing
Arthur Cover and Deep beat a
creates “Turing
Samuel Peter Hart Blue profession
Test” to
coined published an defeated al Go
determine if a
the term article about Garry player
computer has
Machine nearest Kasparo
real intelligence
Learning neighbour v in
algorithm chess
How Machine Learning works?
Why Machine Learning?

Machine Learning is a crucial tool in the process of getting insights from massive
datasets held by companies and researchers today.
There are two main reasons for this:
• Scale of data
• Unexpected findings
Why Machine Learning?

There are two main reasons for this:

• Scale of data
• Companies have massive volumes and varieties of data that need to be
processed.
• Models can be programmed to process data on their own, determine
conclusions, and identify patterns.
• Unexpected findings
• Machine learning (ML) algorithms’ analytical accuracy improves with each
iteration from the datasets it analyses.
• ML uncovers hidden insights without being specifically programmed to do so.
Why Machine Learning?

• The self-driving car.

• Online recommendations, such as those in Amazon and Netflix.

• Knowing what customers are saying about you on Twitter.

• Fraud detection.

These are some widely publicized machine learning applications because

the models iteratively decide when exposed to data.
Machine Learning Use Cases

• Manufacturing: Predictive maintenance and condition monitoring.

• Retail: Upselling and cross-channel marketing.

• Healthcare and life sciences: Disease identification and risk reduction.

• Travel and hospitality: Dynamic pricing.

• Financial services: Risk analytics and regulation.

• Energy: Energy demand and supply optimization .

When to use Machine Learning?

Machine learning scenarios often have the following common properties:

• They involve a repeated decision or evaluation which you want to automate and need
consistent results.

• It is challenging and often impossible to explicitly describe the solution or criteria behind
a decision.

• You require labelled data or existing examples, where you can describe the situation and
map it to the correct result.
What is an Algorithm?
Time for Quiz

21
Open Quiz

Machine Learning is a class of __________ methods.

✔ Statistical
Open Quiz

Machine learning uses large amounts of data and algorithms to perform tasks.

❑ True
❑ False
Open Quiz

Arthur Samuel coined the term Machine Learning in which year?

❑ 1950
❑ 1956
❑ 1958
❑ 1959
Key Takeaways

1. Machine Learning enabled computers and systems to learn

without being explicitly programmed.
2. Machine Learning is a crucial tool used by industry and
researchers to get insights from data.
3. Machine Learning and Traditional Programming have different
approach to solve problems.
What’s next?

26
Reflections

1. ML gives computers the ability to learn without explicit programming. Are

there different ways to classify Machine learning methods?
2. Are there any challenges faced by ML?
Let’s start by understanding what
Machine Learning Models are.

28
Machine Learning Models
Machine Learning Models

• ML Terminologies

• Machine Learning Process

• What are different Machine Learning Models?

• What are the challenges faced by Machine Learning?

Machine Learning Terminologies

Key Machine Learning Terminologies:

• Model
• Feature
• Feature Vector
• Training
• Prediction
• Target (Label)
• Overfitting
• Underfitting
Machine Learning Terminologies
• Model

• A machine learning model is a file

that has been trained to recognize
certain types of patterns.

• E.g., Given a dataset with two

variables, age (input) and height
(output), we can implement a model
to predict the height of a person
based on their age.
Machine Learning Terminologies
• Feature

• A feature is a measurable property or parameter of the data-set.

• E.g., In email spam detector, the features could include the following:
• words in the email text.
• sender's address.
• time of day, the email was sent.
• email contains the phrase "one weird trick."
Machine Learning Terminologies
• Feature Vector

• Feature vectors are used to represent numeric or symbolic characteristics,

called features, of an object in a mathematical, easily analyzable way.

• A numerical representation of objects helps algorithms to do processing and

statistical analysis

• E.g.,

• A color can be described by combination of red, blue, and green

• A feature vector for this would be color = [R, G, B].

Machine Learning Terminologies
• Training
• An ML algorithm takes a set of data known as “training data” as input.
• The learning algorithm finds patterns in the input data and trains the model for
expected results (target).
• The output of the training process is the machine learning model.
• Prediction
• “Prediction” refers to the output of an algorithm after it has been trained on a
historical dataset and applied to new data when forecasting the likelihood of a
particular outcome.
• E.g., whether or not a transaction that already occurred was fraudulent.
• Target (Label)
• The value that the machine learning model must predict, is called the target or label.
Machine Learning Terminologies

• Overfitting
• When a massive amount of data trains a machine learning model, it tends to learn from
the noise and inaccurate data entries.

• The model fails to characterize the data correctly.

• Underfitting
• The model is unable to capture the relationship between the input and output variables
accurately, generating a high error rate on both the training set and unseen data.

• It occurs when a model is too simple, which can be a result of a model needing more
training time, more input features, or less regularization.
 Now that we understand key
ML Terminologies, let's dive deeper Into the
Machine Learning Process

37
Machine Learning Process

Seven Steps of Machine Learning

1. Gathering Data
2. Preparing data
3. Choosing a model
4. Training
5. Evaluation
6. Hyperparameter Tuning
7. Prediction
Let’s try to understand it with the help of
this video

39
Machine Learning Process
Machine learning models are defined
by the presence or absence of human
influence on raw data — whether a
reward is offered, specific feedback is
given, or labels are used.

41
What are different Machine Learning Models?

Machine Learning algorithms fall roughly in three categories:

• Supervised Learning

• Unsupervised Learning

• Reinforcement Learning
What are different Machine Learning Models?
Supervised and Unsupervised Learning
What are the challenges faced by Machine Learning?
• AI impact on jobs
• Public perception around artificial intelligence centers around job loss, this
concern should be probably reframed.

• Privacy
• Data privacy, data protection and data security, are of prime concerns and have
allowed policymakers to make more strides here in recent years.
What are the challenges faced by Machine Learning?

• Bias and discrimination

• Instances of bias and discrimination across a few intelligent systems have
raised many ethical questions regarding the use of artificial intelligence.

• Accountability
• There isn’t significant legislation to regulate AI practices, there is no real
enforcement mechanism to ensure that ethical AI is practiced.
Models for machine learning. (n.d.). IBM Developer. Retrieved September 28, 2021, from https://developer.ibm.com/articles/cc-models-machine-learning/
Discussion

Have you heard about technological singularity? If not refer to:

https://mitpress.mit.edu/books/technological-singularity

How is technological singularity a challenge for Machine Learning?

Key Takeaways

1. Machine Learning is a seven steps process.

2. Machine Learning models are classified into three categories based on

whether a reward is offered, specific feedback is given, or labels are used

3. Machine Learning faces many challenges including privacy.

What’s next?

49
Reflections

1. What are these Machine Learning models? How they work?

2. How Supervised Learning impacts ML algorithm?
Let’s start by understanding what is Supervised Learning.

51
Introduction to Supervised Learning
Introduction to Supervised Learning

• What is Supervised Learning?

• How it works?

• What are different Supervised Learning problems?

• Supervised Learning Algorithm.

• Applications[hands-on activity/discussion].

• Challenges of Supervised Learning.

What is Supervised Learning?

• Supervised learning, is a subcategory of machine learning and artificial

intelligence.

• Goal: To make accurate predictions for new, never-before-seen data.

• Data is labelled.
• It is used whenever we want to predict a certain outcome from a given input, and we
have examples of input/output pairs.

• Supervised Learning requires human effort to build the training set.

• Supervised learning helps organizations solve a variety of real-world problems
at scale, such as classifying spam in a separate folder from your inbox.
How Supervised Learning works?
What are different Supervised Learning problems?
• Two major types of Supervised Learning Problems
• Classification
• Classification maps input to output labels. Classification is a predictive modeling
technique where a label is predicted for a given input.
• E.g., filtering emails to determine whether they are spam or not spam.
• Regression
• Regression maps input to a continuous output. Regression learning allows us to
predict continuous outcome variables based on the value of one or more
predictor variables.
• E.g., Determine the relationship between the number of rash driving cases to the
number of road accident cases by a driver.
What is Supervised Learning?

Classification Problem:
• Binary Classification.
• Distinguishing exactly between two classes.
• E.g., Classifying emails as either spam or not.
• Multiclass Classification
• Classification between more than two classes.
• E.g., Predicting language of a given text.
What is Supervised Learning?
• Regression
• The goal is to predict a continuous number, or real number in mathematical
terms.
• E.g., Predicting a person’s annual income from
• their education,
• their age, and
• where they live.
Supervised Learning Algorithms?

Most used Supervised Learning methods are:

• Neural networks
• Naive Bayes
• Linear regression
• Logistic regression
• Support vector machine (SVM)
• K-nearest neighbor
• Random forest
Supervised Learning Applications

Some common applications include:

• Image- and object-recognition

• Predictive analytics

• Customer sentiment analysis

• Logistic regression

• Spam detection
Challenges in Supervised Learning

• Good examples need to be used to train the data.

• Computation time is very high for Supervised Learning.
• Unwanted data could reduce the accuracy.
• Pre-Processing of data is always a challenge.
• If the dataset is incorrect, you make your algorithm learn incorrectly
which can bring losses.
Time for Quiz

62
Open Quiz

__________ Learning is used whenever we want to predict a certain outcome

from a given input, and we have examples of input/output pairs.

✔ Supervised Learning

63
Open Quiz

If there is a certain type of continuity in the output, which type of problem it is?

❑ Regression
❑ Classification

64
Open Quiz

Machine learning uses large amounts of data and algorithms to perform the
task.

❑ True
❑ False

65
Key Takeaways

1. Supervised Learning subcategory of AI predicts outcome based on

labelled data.

2. Classification and Regression are two different supervised learning.

Reflections

1. Supervised Learning is is used in scenarios where the input, as well as

corresponding outputs, are known. Is there any other method to predict when
outputs are not known?

2. How does Unsupervised Learning predict data?

67
Let’s start by understanding what is
Unsupervised Learning.
Introduction to Unsupervised
Learning
Introduction to Unsupervised Learning

• What is Unsupervised Learning?

• How does it work?

• What are different types of Unsupervised Learning?

• Unsupervised Learning Algorithm.

• Applications[hands-on activity/discussion].

• Challenges in Unsupervised Learning.

70
What is Unsupervised Learning?

• Unsupervised learning, uses machine learning algorithms to analyze and

cluster unlabeled datasets.
• Unsupervised Learning algorithms discover hidden patterns or data groupings
without the need for human intervention.
• Its ability to discover similarities and differences in information make it an ideal
solution for:
• exploratory data analysis.
• cross-selling strategies.
• customer segmentation.
• image recognition.

71
How Unsupervised Learning works?

72
What are different types of Unsupervised Learning?

● Two major types of Unsupervised Learning:

○ Transformations of the dataset.
○ Clustering.

73
What are different types of Unsupervised Learning?

• Transformations of the dataset

• Algorithms that create a new
representation of the data which might
be easier for human or other machine
learning algorithms to understand
compared to the original
representation of the data.
• Common application is Dimensionality
reduction.
• E.g., Representing table in 2 D graph
for visualization.

74
What are different types of Unsupervised Learning?
• Clustering
• Partitioning data into distinct groups of similar items.
• Clustering is the most common unsupervised learning technique.
• It is used for exploratory data analysis to find hidden patterns or groupings in
data.
• Applications for cluster analysis include gene sequence analysis, market
research, and object recognition.

75
Unsupervised Learning Algorithms

• K-means Clustering

• Hierarchical Clustering

• Anomaly Detection

• Principal Component Analysis

• Apriori Algorithm

76
Unsupervised Learning Applications

• News Sections
• Google News uses unsupervised learning to categorize articles on the
same story from various online news outlets.
• Computer vision
• Unsupervised learning algorithms are used for visual perception tasks,
such as object recognition.
• Anomaly detection
• Unsupervised learning models can comb through large amounts of data
and discover atypical data points within a dataset.
• These anomalies can raise awareness around faulty equipment, human
error, or breaches in security.
77
Unsupervised Learning Applications

● Medical imaging
○ Unsupervised machine learning provides essential features to medical
imaging devices, such as image detection, classification and
segmentation, used in radiology and pathology to diagnose patients
quickly and accurately.
● Customer personas
○ Unsupervised learning allows businesses to build better buyer persona
profiles, enabling organizations to align their product messaging more
appropriately.
● Recommendation Engines
○ Using past purchase behavior data, unsupervised learning can help to
discover data trends that can be used to develop more effective cross-
selling strategies.
78
Challenges in Unsupervised Learning

• There is no way of obtaining the method of data sorting, as the dataset is

unlabeled.
• They may be less accurate as the input data is not known and labelled by the
humans making the machine do it.
• The information obtained by the algorithm may not always correspond to the
required output class.
• The user must understand and map the output obtained with the
corresponding labels.
Supervised vs Unsupervised Learning

Supervised Unsupervised

Deals with labelled data Deals with unlabelled data

Useful for solving problems that rely on Useful for finding unknown patterns
past trends for example the price of within data from an experimental
an item. device.

Computing power required is simpler. Computing power required is larger.

Time for Quiz

81
Open Quiz

1. __________ Learning deals with unlabeled data.

✔ Unsupervised Learning

82
Open Quiz

Dimensionality Reduction is common example of?

❑ Transformations of the dataset

❑ Clustering

83
Open Quiz

Unsupervised learning, uses machine learning algorithms to analyze and

cluster unlabeled datasets.

❑ True
❑ False

84
Key Takeaways

1. Unsupervised learning, uses machine learning algorithms to analyze and

cluster unlabeled datasets.

2. Transformations of the dataset and Clustering are two types of

Unsupervised Learning.
What’s next?

86
Reflections

• Supervised Learning learned from Labeled data while Unsupervised Learning

learned from unlabeled data, is their any other way of learning?

87
Now let's get started with
Reinforcement Learning!
Introduction to Reinforcement
Learning
Introduction to Reinforcement Learning

• What is Reinforcement Learning?

• How does it work?

• What are different Reinforcement Learning problems?

• Reinforcement Learning Algorithm.

• Challenges of Reinforcement Learning

90
What is Reinforcement Learning?

91
Let's take a step back and
understand this with an analogy !
Rat In a Maze

● In this case, cheese is the reward, rat is

the agent and maze is the environment.
● The rat must choose the path which
leads to cheese.
● The rat must remember every wrong
path it takes, so that it doesn’t go there
again.
● This helps the rat to move closer and
closer to the reward.
93
Discussion Key Terms in Reinforcement Learning.

• State

• Agent

• Action

• Reward

• Punishment

What do you think that each of these terms depicts in Rat in a Maze game?

94
Let us understand these term with our analogy:

● State: This term signifies the current position of the rat in the maze. It can be
any random position in the maze.
● Agent: Our agent over here is the rat itself.
● Action: The action rat must take to move in the correct direction of the
cheese.
● Reward: The reward is delicious cheese.
● Punishment: The punishment can be the rat not getting the cheese at all.
● Environment: Maze with random paths.

95
Let's try to understand how this works with the help
of Pac-man!

96
Now as we have an idea about
Reinforcement Learning, lets dive a
bit deeper and understand
Q-Learning- a model free
Reinforcement Learning Algorithm.
What is Q – Learning ?

• Q-Learning is a Model free form of Reinforcement Learning. It basically learns

the quality of the action previously taken.

• The Q-learning algorithm can now smartly decide what action to take in which
situation.

• Using our analogy we can say, now the rat just got smarter and now can figure
out the location of the cheese in any maze.

98
How is Q-Learning different from Reinforcement
Learning?

• Quality learning works on similar principles of reinforcement learning, but

additionally it records the action to reward ratio. It is a greedy form of
reinforcement learning.

• It notes what actions are going to maximize the reward and what actions are not
helping to fetch the reward. This is the unique part of Q-Learning.

• If the rat uses Q-learning to fetch the cheese, it will choose the shortest
path ,which will result in getting the cheese in the shortest time.

99
Discussion

• Mention some real-world applications of reinforcement learning?

100
Key Takeaways

1. Reinforcement Learning learn from its own experiences by trial and

error.
2. Q-Learning is a Model free form of Reinforcement Learning.
What’s next?

102
Let’s understanding how ML
differs from Deep Learning and
AI
AI vs ML vs DL
Artificial Intelligence Vs Machine Learning Vs Deep
Learning

• What is Deep Learning?

• What is Neural Network Architecture?

• AI Vs ML Vs DL.

• Why DL?

• DL Applications.

• What are different Reinforcement Learning problems?

• Applications[hands-on activity/discussion].

• Challenges.
105
What is Deep Learning?

• Deep learning is a type of machine learning in which a model learns to

perform classification tasks directly from images, text, or sound.

• Deep learning is usually implemented using a neural network architecture.

• The term “deep” refers to the number of layers in the network—the more
layers, the deeper the network.

• Traditional neural networks contain only 2 or 3 layers, while deep networks

can have hundreds.

106
What is Neural Network?

• Neural networks are loosely modelled after how neurons in the brain
behave.

• They can automatically extract features without input from the

programmer.

• Every neural network node is essentially a machine learning algorithm.

• It is useful when solving problems for which the dataset is very large.

107
What is Neural Network Architecture?

108
Artificial Intelligence Vs Machine Learning Vs Deep
Learning

AI - Any technique that enables computers

to mimic human intelligence.
Machine Learning - A subset of AI that
enables machines to improve at tasks with
experience.
Deep Learning - A subset of ML that
enables software to train itself to perform
tasks with vast amounts of data

109
Machine Learning vs DL?

110
 ML vs DL

How Classic ML works?

• Feature Engineering

• Identify the set of features that uniquely represent a given face

• E.g., the roundness of the face or the distance between the eyes.

• Apply a machine learning classifier algorithm, which learns to associate a

given pattern of features with a unique identity.

111
 ML vs DL

How DL works?

• Deep Learning avoids feature engineering.

• Given enough “labelled data” (i.e., images of known faces) and the right
tuning.

• Deep Learning model will identify the most relevant features from the data
on its own.

112
Deep Learning Applications

• Deep learning is especially well-suited to identification applications, such as

face recognition, text translation, voice recognition, and advanced driver
assistance systems, including, lane classification and traffic sign recognition.

• A self-driving vehicle slows down as it approaches a pedestrian crosswalk.

• An ATM rejects a counterfeit bank note.

• A smartphone app gives an instant translation of a foreign street sign.

113
 Deep learning represents a conceptual
shift in thinking, from
"How do you engineer the best features?"
to
"How do you guide the model toward
discovering the best features?”
114
Bibliography
• Education, I. C. (2021, August 12). Machine Learning. IBM.
https://www.ibm.com/cloud/learn/machine-learning?mhsrc=ibmsearch_a&mhq=machine%20learning
• Label Your Data. (2020, July 6). The History of Machine Learning: How Did It All Start? Label Your Data Team.
https://labelyourdata.com/articles/history-of-machine-learning-how-did-it-all-start
• The Difference Between Artificial Intelligence, Machine Learning and Deep Learning. (n.d.). Intel. Retrieved September 28, 2021, from
https://www.intel.com/content/www/us/en/artificial-intelligence/posts/difference-between-ai-machine-learning-deep-learning.html
• Parthasarathy, S. (2019, April 22). Difference between Traditional programming versus Machine Learning from a PM perspective.
Medium.
https://productcoalition.com/difference-between-traditional-programming-versus-machine-learning-from-a-pm-perspective-3802b02bc7f
6
• GeeksforGeeks. (2020, April 11). Top 10 Apps Using Machine Learning in 2020.
https://www.geeksforgeeks.org/top-10-apps-using-machine-learning-in-2020/
• Replika. (n.d.). Replika.Com. Retrieved September 30, 2021, from https://replika.ai/
• Video Source: What is Machine Learning? (2018, February 22). [Video]. YouTube. https://www.youtube.com/watch?v=QghjaS0WQQU
• What Is Machine Learning? - I School Online. (2021, January 29). UCB-UMT.
https://ischoolonline.berkeley.edu/blog/what-is-machine-learning/
• N. (n.d.). What Is Machine Learning - ML - and Why Is It Important? | NetApp. NetApp. Retrieved September 30, 2021, from
https://www.netapp.com/artificial-intelligence/what-is-machine-learning/
• What is a machine learning model? (2021, May 24). Microsoft Docs.
https://docs.microsoft.com/en-us/windows/ai/windows-ml/what-is-a-machine-learning-model
115
Bibliography
• Framing: Key ML Terminology | Machine Learning Crash Course. (n.d.). Google Developers. Retrieved September 28, 2021, from
https://developers.google.com/machine-learning/crash-course/framing/ml-terminology
• DataRobot. (2021, September 27). Prediction | DataRobot Artificial Intelligence Wiki. DataRobot | AI Cloud.
https://www.datarobot.com/wiki/prediction/
• Education, I. C. (2021a, March 25). Underfitting. IBM. https://www.ibm.com/cloud/learn/underfitting
• Advani, V. (2021, June 10). What is Machine Learning? How Machine Learning Works and future of it? GreatLearning Blog: Free Resources
What Matters to Shape Your Career! https://www.mygreatlearning.com/blog/what-is-machine-learning/
• Advani, V. (2021, June 10). What is Machine Learning? How Machine Learning Works and future of it? GreatLearning Blog: Free Resources
What Matters to Shape Your Career! https://www.mygreatlearning.com/blog/what-is-machine-learning/
• Models for machine learning. (n.d.). IBM Developer. Retrieved September 28, 2021, from
https://developer.ibm.com/articles/cc-models-machine-learning/
• What Is Machine Learning? | How It Works, Techniques & Applications. (n.d.). Mathworks. Retrieved September 29, 2021, from
https://in.mathworks.com/discovery/machine-learning.html
• Models for machine learning. (n.d.). IBM Developer. Retrieved September 28, 2021, from
https://developer.ibm.com/articles/cc-models-machine-learning/
• Muller, A. C., & Guido, S. (2017). Introduction to Machine Learning with Python (First Edition) [E-book]. O’Reilly Media,Inc.
• C. (n.d.-a). Supervised And Unsupervised Machine Learning. Finxter. Retrieved October 1, 2021, from
https://blog.finxter.com/supervised-and-unsupervised-machine-learning/
• Video Source: Object recognition- https://www.youtube.com/watch?v=i0yqhHKWY0A
• Spam detection: https://www.youtube.com/watch?v=_iOxylrN4Io
116