ARTIFICIAL

INTELLIGENCE,
MACHINE LEARNING,
AND DEEP LEARNING
UPDATED IN: APRIL, 2022
BY: CELESTE NG
DIRECT QUOTE FROM SOURCES:
AS INDICATED IN EACH INDIVIDUAL SLIDE
 CONTENTS

1. Artificial 2. Machine 3. Deep

Intelligence Learning Learning 4. Machine
• History & timeline • How machine • Examples Learning
• Branches learns • Unique Applications
• Relationships (AI, • Machine learning requirements
ML, DL) process • Frameworks
• Types of learning
 CONTENTS

1. Artificial 2. Machine 3. Deep

Intelligence Learning Learning 4. Machine
• History & timeline • How machine • Examples Learning
• Branches learns • Unique Applications
• Relationships (AI, • Machine learning requirements
ML, DL) process • Frameworks
• Types of learning
1. ARTIFICIAL INTELLIGENCE
(AI)
AI is a technique which enables machine to mimic human
behaviors and mind
(AI 是一種技術使機器能夠模仿人類行為和思維 )

Source: https://www.youtube.com/watch?v=WSbgixdC9g8&t=19s
 1. AI TIMELINE ( 時間線 )
1952–
1956: The 1974– 1987–
birth of 1980: The 1993: The
artificial first AI second AI
intelligence winter winter

1956– 1980– 1993–2011:

1974: The 1987: AI
golden Boom
years (expert
黃金歲月 system)
繁榮
Dartmouth Workshop of
1956 was organized by
Marvin Minsky, John
McCarthy

Source: 2021, Wikipedia, URL:

1. AI TIMELINE ( 時間線 )
• 1952–1956: The birth of artificial intelligence (symbolics reasoning
– based on rules similar to IF-THEN statements)
• 1956–1974: The golden years (natural language, robotics)
• 1974–1980: The first AI winter (end of funding, limit in computer
speed, exp computer memory )
• 1980–1987: Boom (expert system)
• 1987–1993: the second AI winter (financial setback  expensive to
maintain, difficult to update, they could not learn, only for special
context)
• 1993–2011: AI (Moore’s
大約每兩年 Law, :machine
集成電路（ learning, BD, DL)
IC ）的晶體管的數量加
倍

Source: 2021, Wikipedia, URL: https://learning.acm.org/techtalks/machinelearning

1. ARTIFICIAL INTELLIGENCE - 人工智能的分支
BRANCHES
CELESTE NG:
ML – a branch of AI where its algorithm allows it to learn to perform a task better through
experience or data sets and to create a model for future prediction; Complex problem
- Unmanned
DL – is a subset of ML where its model mimics the structure of the neuron network ofvehicle
our
brain cells - Automated planning

NLP
Speech recognition
- Works along with ML and DL Machine vision
- Works with NLP
- To form a complete human- - Is used in industrial
- To resolve ambiguity in
computer interactions automation & robotics
language
- Rely on digital sensor
in camera to acquire
images

Source: https://www.javatpoint.com/subsets-of-ai
COMPUTER VISION – EXAMPLE-1

A fill-level inspection system at a brewery • Each bottle of beer passes through

an inspection sensor
•  triggers a vision system
•  to flash a strobe light and
take a picture of the bottle.
• After acquiring the image and
storing it in memory, vision
software
•  processes or analyzes it
and
•  issues a pass-fail response
based on the fill level of the
bottle.
• “fail”—it signals a diverter to
reject the bottle.
• On a display:
• An operator can view
rejected bottles and ongoing
process statistics. 
Source:
COMPUTER VISION – EXAMPLE-2
An Intelligent Fabric Defect Detection System (install on weaving
machines)

Requirements:
high power LED light + high
resolution camera + AI + BDA + DL

Source: https://www.youtube.com/watch?v=7zg5OBsySMA
1. AI, MACHINE LEARNING & DEEP
LEARNING – TIMELINE ( 時間線 )

多層的神經網絡
 CONTENTS

1. Artificial 2. Machine 3. Deep

Intelligence Learning Learning 4. Machine
• History & timeline • How machine • Examples Learning
• Branches learns • Unique Applications
• Relationships (AI, • Machine learning requirements
ML, DL) process • Frameworks
• Types of learning
2. MACHINE LEARNING
Machine learning is
• a subset of AI technique, which uses statistical
methods to enable machines to improve with
experience … when exposes to new data
( 機器學習是  AI 技術的一個子集，它使用統計方法使
機器能夠通過經驗進行改進 )

Applications: recommendation systems, face

verification

Source: https://www.youtube.com/watch?v=WSbgixdC9g8&t=19s
2. HOW CAN MACHINES LEARN?
Celeste: the main problems
with dataset are:
Representative? Bias?
Quality? Transparency?
In order to “educate” the machine,
you need these 3 components:
• (1) Datasets. 
• Machine learning systems are
trained on special collections of
samples called datasets.
• The samples can include
numbers, images, texts or any
other kind of data.
• It usually takes a lot of time and
effort to create a good dataset.

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learn
2. HOW CAN MACHINES LEARN? – DATASET
Computer Vision Datasets
xView: xView is one of the most
massive publicly available datasets
of overhead imagery. It contains
images from complex scenes around
the world, annotated using bounding
boxes.
ImageNet: The largest image dataset
for computer vision. It provides an
accessible image database that is
organized hierarchically, according
to WordNet.
Kinetics-700: A large-scale dataset
of video URLs from Youtube.
Including human-centered actions. It
contains over 700,000 videos.
Google’s Open Images: A vast
dataset from Google AI containing
over 10 million images.

Source:
https://pub.towardsai.net/best-datasets-for-machine-learning-data-science-computer-vision-nlp-a
2. HOW CAN MACHINES LEARN? – DATASET
Sentiment Analysis Datasets
Lexicoder Sentiment Dictionary: This dataset
is specific for sentiment analysis. The dataset
Self-driving
contains over 3000 (Autonomous
negative words Driving) Datasets
and over
2000Waymo
positive Open
Clinical Dataset: This
words. is a fantastic
Datasets
sentiment
IMDB dataset resource
reviews: from the folks
MaskedFace-Net: MaskedFace-Net
An interesting at Waymo.
dataset with is a real
overIncludes a vastcontaining
50,000 dataset
movie dataset of
reviews autonomous
human
from driving,
faces with
Kaggle. correct and
enough
Stanford to train deep
incorrectly
Sentiment netsmasks.
worn
Treebank: from zero.
It contains over 137k
Standard
Berkeley
sentiment DeepDrive
images
dataset which
with BDD100k: One
are based
sentiment on the  of the
annotations.
largest
Twitter USdatasets for
Flick-Faces-HQ
Airline self-driving cars,data
dataset [21].
Sentiment: Twitter containing
For more details
on
over 2000about the
hours dataset and
of driving its uses, please visit the 
US airlines from February 2015,experiences
classified asacross
documentation
Newnegative,
York and and
California.on Github.
positive, neutral tweets
COVID-19
Bosch Small Traffic Dataset: The AllenDataset
Light Dataset: Institute of AI
for smallresearch
traffic has released
lights a vast
for deep research dataset of
learning.
over 45,000 scholarly articles about COVID-19.
MIMIC-III: Openly available dataset developed
by the MIT Lab for Computational Physiology,
comprising de-identified health data associated
with ~40,000 critical care patients. It includes
demographics, vital signs, laboratory tests,
medications, and more.
Source:
2. HOW CAN MACHINES LEARN? – DATASET

Source: https://developers.google.com/machine-learning/data-prep
2. HOW CAN MACHINES LEARN? – FEATURES
(2) Features. 
• Are important – the key to the solution of the
task.
Celeste: Example of the
features in “housing price
prediction”: location,
house size, # of bedrooms,
# of bathrooms

• e.g. -- Reinforcement Learning in news recommendation

• User preferences can change frequently, therefore recommending news to users based
on reviews and likes could become obsolete quickly.  RL system can track the
reader’s return behaviors. 
• Construction of such a system would involve obtaining:
• News features: the content, headline, and publisher.
• Reader features: the reader interacts with the content  clicks and shares.
• Context features: timing and freshness of the news
•  “Reward” is … based on these user behaviors.  Source:
https://neptune.ai/blog/reinforcement-learning
-applications

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-difference-dccce18efe7f
2. HOW CAN MACHINES LEARN? –
ALGORITHM

(3) Algorithm. 
• It is possible to solve the same task
using different algorithms.
• Depending on the algorithm, the
accuracy or speed of getting the
results can be different.
• Sometimes in order to achieve better
performance, you combine different
algorithms, like in ensemble learning.

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learn
 2. Steps in Machine Learning Process
Celeste: Identify
your ML problem Steps taken by a
first, then … data scientist in
building a
machine learning
applications 數據
科學家在構建機
器學習應用時採
取的步驟
1. Identify data set 確定數據集
2. Select ML algorithm 選擇 ML 算法
3. Develop analytical model 開發分析模型
4. Train model 訓練模型
5. Generate scores 計算分數

Source: https://www.techemergence.com/artificial-intelligence-industry-an-overview-by-segment/
2. Machine learning algorithms/types (1)
(URL: https://www.analyticsvidhya.com/blog/2017/09/common-machine-learning-algorithms/ )

In the problem
Broadly, there are 3 types of Machine Learning of housing price
Algorithms or Types .. prediction:
• Outcome
1. Supervised Learning 監督學習 variable?
• How it works:  • Independent
– This algorithm consist of a target / outcome variable (or variables
dependent variable) 目標 / 結果變量（或因變量） which is to
be predicted from a given set of predictors (independent
variables) 一組給定的預測變量（獨立變量 ). 
– The training process continues until the model achieves a
desired level of accuracy on the training data.
• Examples of Supervised Learning (analytical model):
Regression 回歸 , Decision Tree, Random Forest隨機森林 ,
KNN, Logistic Regression 邏輯回歸 etc.
2. Machine learning algorithms/types (1) –
(URL: https://www.analyticsvidhya.com/blog/2017/09/common-machine-learning-algorithms/ )

Coding of “supervised
learning” – using the
algorithm of “decision
tree”
2. Machine learning algorithms/types (2)
(URL: https://www.analyticsvidhya.com/blog/2017/09/common-machine-learning-algorithms/ )

2. Unsupervised Learning 無監督學習

• How it works: In this algorithm, we do not have any target or outcome variable to
predict / estimate 沒有任何目標或結果變量來預測 / 估計 .  
• It is used for clustering population 用於群集人口 in different groups, which is widely
used for segmenting customers 細分客戶 in different groups for specific intervention.
• Examples of Unsupervised Learning (analytical model): Apriori algorithm, K-means.
 
3. Reinforcement Learning 強化學習 :
• How it works:  Using this algorithm, the machine is trained to make specific decisions.
• It works this way: the machine is exposed to an environment where it trains itself
continually using trial and error 通過試驗和錯誤不斷訓練自己 .
• This machine learns from past experience and tries to capture the best possible
knowledge to make accurate business decisions 捕捉最佳可能的知識以做出準確的
業務決策 .
• Example of Reinforcement Learning (analytical model): Markov Decision Process 馬
爾科夫決策過程
2. TYPES OF MACHINE LEARNING – NATURE &
OBJECTIVE Learning
Style/ Type

Source:
https://www.oreilly.com
/radar/reinforcement-lea
rning-explained/

識別集群

supervised learning: is to unsupervised reinforcement learning: is to find a

find a correct action for learning: is to find suitable action model that would
performing a task similarities and maximize the total cumulative
differences between reward of the agent
Source:
data points 以最大化 Agent 的總累積獎
https://towardsdatascience.com/reinforc
ement-learning-101-e24b50e1d292 勵 ， 尋找合適的行為模型
2. TYPES OF MACHINE LEARNING –
REQUIREMENTS

3D” 賽車模擬器” - 是採用強化機器學習

Source: https://neptune.ai/blog/reinforcement-learning-applications

Source: https://www.analyticsvidhya.com/machine-learning/types_of_ml/
2. SUPERVISED MACHINE LEARNING

Types:

Problems related to:

classification 分類

Source: https://vas3k.com/blog/machine_learning/?ref=hn
2. EXAMPLE OF SUPERVISED LEARNING TECHNIQUE
Algorithm

Equation has the form Y= a + bX

a= 方程的截距
Celeste: b=coefficient 係數
Inputs +
label-outputs

Source:
https://towardsdatascience.com/simple-machine-learning-model-in-python-in-5-lines-of-code-fe
2. EXAMPLE OF SUPERVISED LEARNING TECHNIQUE

Identify the BEST

“fitted line” for the
model

Features or “x”es
(input values)

Predicted value
of y (=
ŷ )
Coefficients of
Source: “x”es

https://towardsdatascience.com/simple-machine-learning-model-in-python-in-5-lines-of-code-fe
UNSUPERVISED MACHINE LEARNING

Types:

識別數據中的序列 / 順序
Problems related to:

通過相似性分類
cluster 聚類 and
classification 分類

找到隱藏性的數據依賴關係

Source: https://vas3k.com/blog/machine_learning/?ref=hn
Supervised vs. Unsupervised Machine Learning

Source:
https://www.researchgate.net/figure/Supervised-learning-and-unsupervised-learning-Supervi
SUPERVISED VS. UNSUPERVISED LEARNING

Clustering

Regression

因變量和一個或多個自變量
之間的線性關係

Source:
https://www.researchgate.net/figure/Examples-of-Supervised-Learning-Linear-Regression-a
2. MACHINE LEARNING
APPLICATIONS - OVERVIEW
找到隱藏性的數據依賴關係

dimensionality
reduction 降維

通過相似性分聚類

Source:
https://www.oreilly.com/library/
view/java-deep-learning/978178
8997454/a8fce962-51dd-4e29-a
7f9-9bf4fd245b1d.xhtml
2. ADDITIONAL: MACHINE LEARNING – TYPE,
ALGORITHM & APPLICATION AREAS

Source: https://www.codeproject.com/Articles/5245488/Introduction-to-Machine-Learning-and-ML-NET-Part-1
2. ADDITIONAL EXAMPLES OF SUPERVISED
LEARNING SYSTEM / APPLICATION AREAS

語音筆錄

Supervised machine learning systems. (Source: Brynjolfsson and McAfee 2017b)

 CONTENTS

1. Artificial 2. Machine 3. Deep

Intelligence Learning Learning 4. Machine
• History & timeline • How machine • Examples Learning
• Branches learns • Unique Applications
• Relationships (AI, • Machine learning requirements
ML, DL) process • Frameworks
• Types of learning
3. DEEP LEARNING

Deep learning is
• A particular kind of ML that
• Is inspired by the functionality of brain cells (neuron),
• [Celeste Ng: … its model mimics the structure of our
neuron network of our brain cells]
• which leads to the concept of artificial neural network
(ANN)
( 一種特殊的機器學習 , 來自腦細胞神經元功能的啟
發， 這導致了人工神經網絡的概念 )
• Learning styles: can be supervised, semi-supervised, 
unsupervised or reinforcement learning (as in AlphaGo Zero)
--- Source: Wikipedia, 2020

Source: https://www.youtube.com/watch?v=WSbgixdC9g8&t=19s
DEEP LEARNING – CONCEPT OF “ARTIFICIAL
NEURAL NETWORK”
Nodes (artificial neurons) cf. brain’s neurons
Edges (connections) cf. brain’s synapses – signal
transmission ( 每個連接，就像生物大腦中的突觸一樣，可以向其他神經
元傳輸信號 )
Layers – consist of nodes / neurons & are meant for
diff. input transformations
Node output = Node activation function
= f (k)
= f (k =Bias + sum_of_its_ inputs)
f = can be a sigmoid function

Edge has = weight (change thru learning)

Applications: face recognition, automated trading

systems, cybersecurity (malware, credit card fraud),
Diagram drawn by: Celeste Ng, 2021
email-filtering
DEEP LEARNING

Deep learning is
• Modeled using layers of artificial neurons or
computational units
• to receive input
• In a simple model
• the first layer is input layer,
• followed by a hidden layer, and
• lastly by an output layer. Source:
https://chatbotslife.com/deep-learning-in-
7-lines-of-code-7879a8ef8cfb
• Each layer contains one or more neurons.

Source: https://towardsdatascience.com/cousins-of-artificial-intelligence-dda4edc27b55
3. DEEP LEARNING – A REAL EXAMPLE

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE: TRAINING
ALGORITHM TO IDENTIFY HAND-WRITTEN DIGIT
# of layers & # of
neurons in each are
Input layer, output arbitrary (to start
layer & hidden layer with)

Each pixel is fed into a

neuron  this forms
the “Input Layer”

Each image is
represented
by 28x28
pixels Each
neuron
represents a
digit

Hidden Layer
Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE
Connected channels, weighted channels,

Each channel has a value

Information is attached to it, thus, it is
transferred from one called “weighted
layer to another channels”
through “connected
channels”

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE
Which neurons are activated? “Bias”
Bias, weighted sum of inputs, activation is added to the “weighted sum of
function inputs ” reaching the neuron, which is
then applied to a function, known as
Each neuron has a “activation function”
unique number
f(k=Bias + Weighted sum of
associated with it,
inputs)
called “bias”(i.e. Weighted
[activation] sum of
inputs

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE
The results of the “activation function”
Activation function results determines which neurons are activated.
The activated neurons passed the
information to the following layers.
… continue until the 2nd last layer

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE
The one neuron activated in the output
Final result layer corresponds to the input digit

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. DEEP LEARNING – A REAL EXAMPLE

Adjustment The weights and bias are continuously adjusted

to produce a “well-trained network”

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
FEATURE & NEURON

Edges Patterns

Digits

Edges 
patterns 
digit

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
PARAMETER FOR THE CONNECTIONS
Weighted sum of inputs
a
Each neuron has a number (0 – 1);
a = activation (a number in a neuron)
w = weight of a channel;
can be positive or negative

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
WEIGHTED SUM
The computed
weighted sum can
The activation function be any negative, 0
or positive numbers

BUT, we want
the activation
value to be
between 0 and 1

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
SIGMOID FUNCTION BUT, we want
the activation
value to be
The activation function between 0 and 1
The common
function to do this is
called “sigmoid”
function or logistic
curve

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
SIGMOID FUNCTION
“… the activation of neuron
here is basically a measure
The activation function of how positive the relevant
weighted sum is

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
BIAS
Bias = help to determine which neuron is activated

a = activation; w = weight

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
TOTAL # OF WEIGHTS & BIAS (= PARAMETERS)

a = activation; w = weight

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
MATRIX VECTOR FORM

a = activation; w = weight

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
MATRIX VECTOR FORM

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
3. ADDITIONAL: DEEP LEARNING – A REAL EXAMPLE:
PROGRAM CODE

Source:
https://medium.com/ai-in-plain-english/artificial-intelligence-vs-machine-learning-vs-deep-learning-whats-the-diffe
rence-dccce18efe7f
3. ADDITIONAL: EXAMPLES OF DEEP LEARNING CODE
•Scikit-learn •Matplotlib List of important
•NuPIC •Theano Python libraries
•Ramp •SymPy
•NumPy •Caffe2
•Pipenv •Seaborn
•TensorFlow •Hebel
•Bob •Chainer
•PyTorch •OpenCV Python
•PyBrain •Theano
•MILK •NLTK
•Keras •SQLAlchemy
•Dash •Bokeh
•Pandas •Important Python Libraries for D
•Scipy Source:
https://www.mygreatlearning.com/blog/open-source-py
thon-libraries/

Source: https://chatbotslife.com/deep-learning-in-7-lines-of-code-7879a8ef8cfb
3. ADDITIONAL: EXAMPLES OF DEEP LEARNING CODE

Source: https://chatbotslife.com/deep-learning-in-7-lines-of-code-7879a8ef8cfb
3. DEEP LEARNING FRAMEWORKS

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS
Read on Your Own (ROYO)

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
3. DEEP LEARNING FRAMEWORKS

Source: https://www.youtube.com/watch?v=_FBXfarXKuA
 CONTENTS

1. Artificial 2. Machine 3. Deep

Intelligence Learning Learning 4. Machine
• History & timeline • How machine • Examples Learning
• Branches learns • Frameworks Applications
• Relationships (AI, • Machine learning
ML, DL) process
• Types of learning
4. DEEP LEARNING – APPLICATIONS

Source: https://www.youtube.com/watch?v=6M5VXKLf4D4
4. DEEP LEARNING – APPLICATIONS

Source: https://www.mygreatlearning.com/blog/deep-learning-applications/
4. MACHINE LEARNING
APPLICATIONS

• 1. Virtual Personal Assistants 虛擬個人助理 NLP, ML – RL, Speech

Recog.
• Siri, Alexa, Google [Assistance] are some of the popular examples of
virtual personal assistants. As the name suggests, they assist in finding
information, when asked over voice.
ML -
• 2. Predictions while Commuting 通勤時的預測 SL
• Traffic Predictions: We all have been using GPS navigation services.
While we do that, our current locations and velocities are being saved at a
central server for managing traffic. 
• Uber use ML to define price surge hours by predicting the rider demand

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
4. MACHINE LEARNING
APPLICATIONS
• 3. Videos Surveillance 視頻監控 Vision

• A difficult job to do and boring as well. This is why the

idea of training computers to do this job makes sense.
• It is possible to detect crime before they happen. They
track unusual behaviour of people like standing
motionless for a long time, stumbling, or napping on
benches etc.

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
 4. MACHINE LEARNING
APPLICATIONS
ML - RL
• 4. Social Media Services
• Personalizing your news feed ( 動態消息 ), providing better ads
targeting
• The applications of ML.
• People You May Know: Machine learning works on a simple
concept: understanding with experiences. Facebook
continuously notices the friends that you connect with, the
profiles that you visit very often, your interests, workplace, or
a group that you share with someone
• Face Recognition: You upload a picture of you with a friend
and Facebook instantly recognizes that friend.
Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
4. MACHINE LEARNING
APPLICATIONS
• 5. Email Spam and Malware Filtering ML - UL
• Over 325, 000 malwares ( 惡意軟件 ) are detected everyday and each piece of code is 90–
98% similar to its previous versions.
• The system security programs that are powered by machine learning understand the
coding pattern.
• Therefore, they detects new malware with 2–10% variation easily and offer protection
against them.
NLP, ML –
• 6. Online Customer Support
RL
• Chatbots tend to extract information from the website and present it to the customers.
• Chatbots advances with time. They tend to understand the user queries better and serve
them with better answers, which is possible due to its machine learning algorithms.

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
4. MACHINE LEARNING
APPLICATIONS
• 7. Search Engine Result Refining 搜索引擎結果優化 ML –
RL
• Google and other search engines use machine learning to improve the
search results for you.
• Every time you execute a search, the algorithms at the backend keep a
watch at how you respond to the results.
• If you open the top results and stay on the web page for long, the search
engine assumes that the results it displayed were in accordance to the
query. 

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
4. MACHINE LEARNING
APPLICATIONS
• 8. Product Recommendations 產品推薦 ML – UL

• You shopped for a product online few days back and then
you keep receiving emails for shopping suggestions.
• If not this, then you might have noticed that the shopping
website or the app recommends you some items that
somehow matches with your taste. Certainly, this refines
the shopping experience

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
4. MACHINE LEARNING
APPLICATIONS
• 9. Online Fraud Detection 在線欺詐檢測 ML –
• Machine learning is proving its potential to make cyberspace a UL

secure place and tracking monetary frauds online is one of its

examples.
• For example: Paypal is using ML for protection against money
laundering ( 打擊洗錢 ).
• The company uses a set of tools that helps them to compare
millions of transactions taking place and distinguish between
legitimate or illegitimate transactions taking place between
the buyers and sellers ( 區分合法或非法交易 )

Source:
https://medium.com/app-affairs/9-applications-of-machine-learning-from-day-to-day-life-11
OTHER REFERENCES
1. https://www.datarobot.com/wiki/model/
2. https://www.analyticsvidhya.com/blog/2017/09/c
ommon-machine-learning-algorithms/

3. https://towardsdatascience.com/coding-deep-lear
ning-for-beginners-types-of-machine-learning-b9
e651e1ed9d

4. https://medium.com/ai-in-plain-english/artificial-
intelligence-vs-machine-learning-vs-deep-learnin
g-whats-the-difference-dccce18efe7f

5. https://www.youtube.com/watch?v=IHZwWFH
Wa-w
GENETIC ALGORITHM Back

Source: Derek Kane, 2020, https://www2.slideshare.net/DerekKane/data-science-part-xiv-genetic-algorithms

ARTIFICIAL INTELLIGENCE – FIVE MAIN
TRIBES / APPROACHES / CATEGORIES
家組 根源
rule-based 哲學
interconnected 反向傳播算法
mechanism
神經科學
evolutionary 基因編程
processes 進化生物學
概率推斷
probabilistic
rules 類比器 內核機器算法
template matching

Symbolists Connectionists Evolutionists

Bayesians Analogizers

Source: Pedro Domingos, 2015, The Five Tribes of Machine Learning

with Pedro Domingos; URL:
2. MACHINE LEARNING – THE
GEOMETRY PROBLEM

Source:
https://www.researchgate.net/figure/Example-of-a-geometr
y-problem-used-for-the-investigation-of-the-target-maps-ap
plication_fig5_252642618

Geometry = about lines and curves

Regression is about finding
the best fit of a “line”
Regression is used as an algorithm in
ML  prediction

Thus, ML is about “solving

geometry problem
2. MACHINE LEARNING – THE
GEOMETRY PROBLEM

Source:
https://www.researchgate.net/figure/Example-of-a-geometr
y-problem-used-for-the-investigation-of-the-target-maps-ap
plication_fig5_252642618

Geometry = about lines and curves

Regression is about finding
the best fit of a “line”
Regression is used as an algorithm in
ML  prediction

Thus, ML is about “solving

geometry problem
3. DEEP LEARNING – ACTIVATION FUNCTION

• There are various kind of

Activation Functions that
exists, … ,
• which can help networks
to converge faster or use
less layers etc.

Source:
https://medium.com/@shrutijadon10104776/survey-on-activation-functions-for-deep-learning-9689331ba09