Chapter 8

Flashcards

To enable a learning agent to adapt to dynamic

environments, make complex decision-making,
improve performance, handle uncertainty,
generalize and transfer learning, and reduce human
intervention in real-time decision-making.

Task (T) is the specific job or problem the machine

learning model is trying to solve, experience (E) is
Identify and describe the task (T), experience
the data that the model learns from, and
(E), and performance(P) associated with
performance (P) is the measure of how well the
machine learning applications.
model performs the task.

Common Uses for Machine Learning include

training models and improving accuracy. The
quality, quantity, and diversity of training data
determine how well the model learns.

The core component of machine learning involves

learning agents, supervised and unsupervised
learning, and reinforcement learning.

Supervised learning is a machine learning approach

that’s defined by its use of labelled datasets to train
or “supervise” algorithms into classifying data or
predicting outcomes accurately.

Supervised learning works by categorizing items

into specific categories, such as classifying images
or predicting outcomes based on labelled datasets.

Medical diagnosis, credit scoring, image

classification, sentiment analysis, handwriting
recognition, and fraud detection are examples of
classification in various domains.
Supervised learning also works by identifying real
values and training models to predict outcomes
based on labelled datasets.

House price prediction, genomic data analysis, and

market basket analysis are examples of regression
in various domains.

Unsupervised learning involves clustering and

dimensionality reduction to discover patterns,
structures, or relationships within unlabeled data.

Reinforcement learning is a subfield of machine

learning where an agent learns to make decisions
by interacting with an environment and learning
through trial and error based on feedback from the
environment.

The applications of reinforcement learning include

safe and ePicient navigation, obeying traPic rules,
avoiding collisions, adapting to traPic conditions,
and minimizing travel time in a self-driving car
environment.

Supervised learning involves training on labelled

data, unsupervised learning involves training on
unlabeled data, and reinforcement learning
involves training a model through interactions with
an environment.

Supervised learning examples include fraud

detection and diagnostics, unsupervised learning
examples include customer segmentation and
image segmentation, and reinforcement learning
examples include finances and self-driving cars.
Smart Notes
Machine Learning

•Introduction to Machine Learning (ML)

Revision: Supervised Vs Unsupervised Vs Reinforcement

•Machine Learning (ML) is a subset of artificial intelligence (AI) that focuses on the
development of algorithms and statistical models that allow computers to perform tasks

without being explicitly programmed for each specific task.

•Definition: ML enables computers to learn from data.

•Why Would We Want an Agent to Learn?

•ML allows for adaptation to dynamic environments, complex decision-making, improved

performance, handling uncertainty, generalization and transfer learning, and handling large

data.

•Learning Agent

•Reduced human intervention and real-time decision-making.

•Machine Learning Applications

•Predict the relevance of web pages, search engines, suggest movies, shows, or music,

provide product recommendations, Facebook's friend tag suggestions, understand user

commands, smart assistants, categorize emails, suggest the fastest routes, and fraud

detection.

•Task, Experience, and Performance Associated with Machine Learning Applications

•Data: The Fuel for ML

•Importance of Data in Machine Learning: Training models and improving accuracy rely on
the quality, quantity, and diversity of training data. The more data a model is trained on, the

better it becomes at making predictions.

Machine Learning and Data Characteristics

•Real-world phenomena are intricate, and a larger dataset captures this complexity, helping

to build models closer to reality.

•Structured Data: organized in rows and columns.

•Unstructured Data: information not organized in a predefined manner.

•Semi-structured Data: a mix of structured and unstructured data.

•Examples of Data Sources: Healthcare, Social Media, Communication, Internet of Things,

Research, Education, Financial, E-commerce, Environmental, Transportation, Entertainment.

•Machine Learning Task for Everyday Data: Predicting stock prices based on financial data,
recommending products on e-commerce platforms, predicting traffic flow based on

transportation data, predicting user preferences on entertainment platforms.

•Importance of Training Data Quality: Accurate, unbiased, and relevant data is critical for

successful machine learning.

•Importance of Training Data Quantity: The amount of training data can significantly

influence an AI system's performance.

•Machine Learning Models: Computer programs used to recognize patterns in data or make
predictions, decisions, or classifications. Models encapsulate the knowledge gained from

training data and are used to make predictions on new, unseen data.
Overview of Supervised Learning in Machine Learning

•Supervised learning involves sorting items into categories (classification) and identifying

real value (regression).

•It uses labelled datasets to train algorithms to classify data or predict outcomes accurately.

•Examples of classification in various domains include medical diagnosis, credit scoring,

image classification, sentiment analysis, handwriting recognition, and fraud detection.

•Regression in various domains involves predicting house prices and temperature.

Applications of Forecasting and Unsupervised Learning in Various Industries

•Forecasting includes applications such as weather forecasting, stock price prediction,

healthcare cost prediction, energy consumption forecasting, and GPA prediction.

•These applications involve predicting future outcomes based on historical data and relevant

features specific to each domain.

•Unsupervised learning techniques such as clustering and dimensionality reduction are used

to analyze and group unlabelled data without human intervention.

•Clustering is used in various domains such as customer segmentation, image segmentation,

social network analysis, and document clustering.

•Each domain has specific features and applications for using clustering techniques to

uncover patterns and relationships within the data.

Types of Machine Learning Algorithms

•Genomic Data Analysis:

•Problem: Group genes with similar expression patterns across samples.

•Features: Gene expression levels under different conditions.

•Application: Understanding gene interactions and identifying biomarkers.

•Market Basket Analysis:

•Problem: Discover associations and patterns in customer purchasing habits.

•Features: Purchased items in transactions

•Application: Retailers use this for product recommendations and store layout optimization.

•Reinforcement Learning:

•Definition: RL is a subfield of machine learning where an agent learns to make decisions by

interacting with an environment through trial and error based on feedback.

•Applications: Environment for a self-driving car, learning objectives, reward design, state

representation, and action space.

•Supervised Learning:

•Definition: Training a model on labeled data to learn a mapping from input features to the

correct output labels.

•Examples: Fraud detection, email spam detection, diagnostics, image classification, house

price prediction, and healthcare cost prediction.

•Unsupervised Learning:

•Definition: Training a model on unlabeled data to discover patterns, structures, or

relationships within the data.

•Examples: Customer segmentation, image segmentation, social network analysis, document

clustering, and dimensionality reduction in genomic data analysis.

•Revision: Examples of Learning Types in Various Domains:

•Examples of machine learning algorithms in different domains such as finance,

manufacturing, stock management, and self-driving cars.

Machine Learning Building Steps

•The process of building effective machine learning solutions involves a series of well-

defined steps from initial problem formulation to successful deployment of functional

models.

•Understanding the problem to solve using machine learning involves identifying the goal

and any limits, as well as determining the type of machine learning needed.

•Data Gathering & Preparing involves identifying various data sources, collecting data,
integrating data from different sources, and preparing it for use in machine learning training

and testing.

•Preparing Data is the step where data is put into a suitable place and prepared to be used

in machine learning.

•Selecting & Training The Model includes choosing an appropriate model, using training

data, and adjusting model parameters for optimal performance.

•Testing & Deploying The Model involves evaluating the trained model's performance and

preparing it for deployment in a production environment.

Summary
"The Fundamentals of Machine Learning: Definitions, Applications, and Building Steps"

This is a summary of a lecture on Machine Learning (ML) as part of an Introduction to

Artificial Intelligence course. The lecture covers the definition of ML, its applications, main
components, and building steps. ML is a way for computers to learn from data without
explicit programming, improving their performance on tasks over time. The importance of

data in ML is emphasized, as well as the role of models in ML.

"Understanding Real-World Data and Training for Machine Learning"

This text discusses various types of real-world data, such as structured, semi-structured, and
unstructured data, and provides examples like healthcare, social media, and financial data. It
also emphasizes the importance of high-quality and quantity of training data for successful
machine learning (ML) tasks. Furthermore, it explains that ML models are the core
components of ML systems that encapsulate knowledge gained from training data and can

make predictions on new data.

"Unlocking the Potential of Supervised Learning in Machine Learning"

Supervised learning is a machine learning approach using labeled datasets to train

algorithms for accurate data classification or prediction. It includes tasks like image

classification, sentiment analysis, and house price prediction.

"Unsupervised Learning: Discovering Hidden Patterns Without Human Intervention"

Unsupervised learning focuses on clustering unlabelled data, discovering hidden patterns

without human intervention. Applications include customer segmentation for targeted
marketing, image segmentation for medical imaging and object detection, social network
analysis for understanding community dynamics, and document clustering for topic

discovery.

"Exploring Insights in Data Analysis and Learning Algorithms"

Genomic data analysis identifies gene interactions and biomarkers, while market basket

analysis discovers customer purchasing patterns for retailers. Reinforcement learning is a

trial-and-error approach to decision-making in dynamic environments, such as self-driving
cars. Supervised learning involves training on labeled data for tasks like classification and

regression, while unsupervised learning explores unlabelled data for clustering and

dimensionality reduction.

"A Comprehensive Guide to the Machine Learning Building Process"

The machine learning building process consists of steps from problem formulation to
deployment, including gathering and preparing data, selecting and training the model,

evaluating performance, and deploying the model in a production environment.

Quiz
‫‪Good luck on your finals J‬‬

‫‪Made by: Rayan MohammadNaeem Tashkandy‬‬

‫ﻻ ﺗ ﻧ ﺳ و ﻧ ﺎ ﻣ ن ﺻ ﺎ ﻟﺢ د ﻋ ﺎ ﺋ ﻛ م ﷲ ﯾ و ﻓ ﻘ ﻛ م ﯾ ﺎ ر ب ا ﻣ ﯾ ن‬
‫ر اﺟ ﻌ و ا اﻟﺳ ﻼ ﯾد ا ت ﻛ ﻣ ﺎ ن ا ﺣ ﺗ ﯾ ﺎ ط‬