Introduction to Machine Learning

Introduction to Machine Learning

Chapter 1
2
C0 - Personal Information
Agenda
 What is machine learning?
 History and relationships to other fields
 Applications of machine learning
 Types of machine learning techniques
 Overview of Data mining and KDD process:
 Prediction vs. Description modeling

C0 - Personal Information
3
AI
 AI is defined as the
intelligence of a
machine or computer
that enables it to
imitate or mimic human
capabilities.
 How it works ->

C0 - Personal Information
4
Machine Learning - ML
 ML is a type of AI that allows software apps to become
more accurate at predicting outcomes without being
explicitly programmed to do so.
 ML provides machines with the ability to automatically
learn from data and past experiences while identifying
patterns to make predictions with minimal human
intervention.
 ML algorithms use historical data as input to predict
new output values.
C0 - Personal Information
5
AI – Terminologies

C0 - Personal Information
6
AI…
 Deep learning
 Subset of ML that learns by processing data with the help
of artificial neural networks.
 Neural Networks
 Computer systems that are loosely modeled on neural
connections in the human brain and enable deep learning.

C0 - Personal Information
7
AI…
 Cognitive computing:
 Aims to recreate the human thought process in a computer model.
 It seeks to imitate and improve the interaction between humans
and machines by understanding human language and the
meaning of images.
 NLP
 A tool that allows computers to comprehend, recognize, interpret,
and produce human language and speech.
 Computer vision:
 Employs deep learning and pattern identification to interpret
image content (graphs, tables, PDF pictures, and videos).
C0 - Personal Information
8
ML…
 Problems solved by ML today
 Computational finance (credit scoring, algorithmic
trading)
 Computer vision (facial recognition, motion tracking,
object detection)
 Computational biology (DNA sequencing, brain tumor
detection, drug discovery)
 Automotive, aerospace, and manufacturing (predictive
maintenance)
 Natural language processing (voice recognition)
C0 - Personal Information
9
ML – How it
works

C0 - Personal Information https://www.spiceworks.com/tech/artificial-intelligence/articles/what-is-ml/ 10

Traditional Program vs ML

C0 - Personal Information
11
 Traditional Programs
 Imagine, the challenging task of writing a program that can
detect if a cat is present in an image.
 Details like varying lighting conditions, different types of
cats, and various poses a cat might be in.
 Machine Learning
 The overall goal is to use a model created by a model
training algorithm to generate predictions or find patterns
in data that can be used to solve a problem.

C0 - Personal Information
12
History of ML
 Read on this …

C0 - Personal Information
13
ML – Relationship to other fields

C0 - Personal Information
14
ML – Types

C0 - Personal Information https://www.spiceworks.com/tech/artificial-intelligence/articles/what-is-ml/ 15

ML - types…
 Supervised learning,
 Every training sample from the dataset has a corresponding
label or output value associated with it.
 As a result, the algorithm learns to predict labels or output
values.
 Classification: These refer to algorithms that address
classification problems where the output variable is categorical.
 yes or no, true or false, male or female,
 Regression: Regression algorithms handle regression problems
where input and output variables have a linear relationship.
 weather prediction, market trend analysis,

C0 - Personal Information
16
ML - types…
 Unsupervised learning,
 There are no labels for the training data.
 A ML algorithm tries to learn the underlying patterns or
distributions that govern the data.
 Clustering: grouping objects into clusters based on
parameters such as similarities or differences between
objects.
 Grouping customers by the products they purchase.
 Association: It determines the dependency of various
data items and maps associated variables.
 Web usage mining and market data analysis. 17
C0 - Personal Information
ML - types…
 Semi-supervised learning
 Comprises characteristics of both supervised and
unsupervised machine learning.
 Drawbacks of supervised learning and unsupervised
learning algorithms
 Student – learning concept under teachers supervision and self
learning, the combine this for the final exam.

C0 - Personal Information
18
ML - types…
 Reinforcement learning,
 The algorithm figures out which actions to take in a
situation to maximize a reward (in the form of a number)
on the way to reaching a specific goal.
 This is a completely different approach than supervised
and unsupervised learning.
 The component is rewarded for each good action and
penalized for every wrong move.
 Unlike supervised learning, reinforcement learning lacks
labeled data, and the agents learn via experiences only.
Consider video games.
C0 - Personal Information
19
ML - types…
 Reinforcement learning lacks labeled data, and the
agents learn via experiences only. Consider video
games.
 Here, the game specifies the environment, and each
move of the reinforcement agent defines its state.
 The agent is entitled to receive feedback via
punishment and rewards, thereby affecting the overall
game score.
 The goal of the agent is to achieve a high score.
C0 - Personal Information
20
Components of ML
 Nearly all tasks solved with ML involve three primary
components:
 A ML model - is a block of code or framework that can
be modified to solve different but related problems
based on the data provided.
 extremely generic program, made specific by the data used
to train it.
 A model training algorithm - how model and data are
used to create a trained model
 A model inference algorithm - Using Your Trained
Model
C0 - Personal Information
21
ML – the 5 steps

C0 - Personal Information
22
ML – the 5 steps…
 Step One: Define the Problem
 Define a very specific task – adding $0.5 on snow cone may increase
sales?
 Identify the ML task we might use to solve this problem - Supervised
learning, Unsupervised learning (The presence or absence of labeling in
your data is often used to identify a machine learning task.)
 Further Classifying by using Label Types
 Categorical
 Continuous

C0 - Personal Information
23
ML – the 5 steps…
 Step Two: Build a Dataset
 The most important step of the machine learning process

C0 - Personal Information
24
ML – the 5 steps… Build a Dataset
 Data inspection
 The quality of your data will ultimately be the largest factor
that affects how well you can expect your model to
perform. As you inspect your data, look for:
 Outliers
 Missing or incomplete values
 Data transformation
 Summary statistics - mean, standard deviation
 Data visualization - see outliers and trends in your
data and to help stakeholders understand your data.
C0 - Personal Information
25
ML – the 5 steps…
 Step Three: Model Training
 Splitting your Dataset
 The model training algorithm iteratively updates a model's
parameters to minimize some loss function.
 Model parameters: settings or configurations the
training algorithm can update to change how the
model behaves.
 Loss function: is used to codify the model’s distance
from this goal.
C0 - Personal Information
26
ML – the 5 steps…
 Putting it All Together – Model Training
 The end-to-end training process is
 Feed the training data into the model.
 Compute the loss function on the results.
 Update the model parameters in a direction that reduces loss.
 You continue to cycle through these steps until you
reach a predefined stop condition.

C0 - Personal Information
27
ML – the 5 steps…
 Step Four: Model Evaluation
 Evaluate how well your model is performing.
 The metrics used for evaluation are likely to be very
specific to the problem you have defined.
 Using Model Accuracy - is a common evaluation metric.
 is the fraction of predictions a model gets right.

C0 - Personal Information
28
ML – the 5 steps…
 Remember: This Process is Iterative

 Every step we have gone through is highly iterative

and can be changed or re-scoped during a project

C0 - Personal Information
29
ML – the 5 steps…
 Step Five: Model Inference
 Congratulations! You're ready to deploy your model.
 Once you have trained your model, have evaluated its
effectiveness, and are satisfied with the results, you're
ready to generate predictions on real-world problems
using unseen data in the field.
 In machine learning, this process is often called inference.

C0 - Personal Information
30
Applications of ML
 Healthcare industry - wearable fitness trackers,
smart health watches – Drug discovery, disease
detection.
 Finance sector – credit scoring
 Retail sector - recommend items based on users’
purchase history
 Travel industry – self driving cars, Ride/uber/feres
 Social media – Advert, recommendation,.
C0 - Personal Information
31
Data Mining
 Is the method of extraction of data or previously
unknown data patterns from huge sets of data.
 We 'Mine for specific data' from the large data set.
 Data mining is also called Knowledge Discovery
Process, is a field of science that is used to determine
the properties of the datasets.
 Gregory Piatetsky-Shapiro founded the term
"Knowledge Discovery in Databases" (KDD) in 1989.

C0 - Personal Information
32
Data Mining vs ML
 Data mining is the probing of available
datasets in order to identify patterns and
anomalies.
 Machine learning is the process of
machines learning from heterogeneous
data in a way that mimics the human
learning process.

C0 - Personal Information
33
AI vs ML vs DL

C0 - Personal Information
34
Thank You!

?C0 - Personal Information

35