MACHINE LEARNING TECHNIQUES

BCDS062

By
Dr Ravi Prakash Verma
Professor
Department of CSAI
ABESIT
 Syllabus – UNIT 1
• INTRODUCTION –
• Learning
• Types of Learning
• Well defined learning problems
• Designing a Learning System
• History of ML
• Introduction of Machine Learning Approaches –
• Artificial Neural Network
• Clustering
• Reinforcement Learning
• Decision Tree Learning
• Bayesian networks
• Support Vector Machine
• Genetic Algorithm
• Issues in Machine Learning
• Data Science Vs Machine Learning.
 Machine Learning
• Machine Learning (ML) is a branch of artificial intelligence (AI) that
enables computers to learn from data and make decisions or
predictions without being explicitly programmed.
• Instead of following a fixed set of rules, ML algorithms identify
patterns in data and improve their performance over time.
• Optimize a performance criterion using example data or past
experience.
• Role of Statistics: Inference from a sample
• Role of Computer science: Efficient algorithms to
• Solve the optimization problem
• Representing and evaluating the model for inference
 Machine Learning
• Machine learning (ML) is a branch of artificial intelligence (AI) focused
on enabling computers and machines to imitate the way that humans
learn, to perform tasks autonomously, and to improve their
performance and accuracy through experience and exposure to more
data.
• UC Berkeley breaks out the learning system of a machine learning
algorithm into three main parts.
• A Decision Process:
• In general, ML algorithms are used to make a prediction or classification.
• Based on some input data, which can be labeled or unlabeled, your algorithm
will produce an estimate about a pattern in the data.
 Machine Learning
• An Error Function:
• An error function evaluates the prediction of the model.
• If there are known examples, an error function can make a comparison to
assess the accuracy of the model.
• A Model Optimization Process:
• If the model can fit better to the data points in the training set, then weights
are adjusted to reduce the discrepancy between the known example and the
model estimate.
• The algorithm will repeat this iterative “evaluate and optimize”
process, updating weights autonomously until a threshold of accuracy
has been met.
 Types of Machine Learning
1.Supervised Learning
1. The model learns from labeled data (input-output pairs).
2. Example: Predicting house prices based on features like size, location, and number
of rooms.
3. Algorithms: Linear Regression, Decision Trees, Support Vector Machines (SVM),
Neural Networks.
2.Unsupervised Learning
1. The model identifies patterns and relationships in data without labeled outputs.
2. Example: Customer segmentation in marketing.
3. Algorithms: K-Means Clustering, Principal Component Analysis (PCA), Autoencoders.
3.Reinforcement Learning
1. The model learns through trial and error by receiving rewards or penalties.
2. Example: AI playing chess or self-driving cars.
3. Algorithms: Q-Learning, Deep Q-Networks (DQN), Policy Gradient Methods.
 Applications of Machine Learning
• Some of the applications are

• Speech Recognition (e.g., Google Assistant, Siri)

• Image Recognition (e.g., Face recognition, object detection)
• Recommendation Systems (e.g., Netflix, YouTube, Amazon)
• Healthcare (e.g., Disease diagnosis, drug discovery)
• Autonomous Vehicles (e.g., Self-driving cars)
 Supervised Learning
• Supervised machine learning, is defined by its use of labeled datasets
to train algorithms to classify data or predict outcomes accurately.
• As input data is fed into the model, the model adjusts its weights until
it has been fitted appropriately.
• This occurs as part of the cross validation process to ensure that the
model avoids overfitting or underfitting.
• Supervised learning helps organizations solve a variety of real-world
problems at scale, such as classifying spam in a separate folder from
your inbox.
 Supervised Learning
• Some methods used in supervised learning include neural networks,
naïve bayes, linear regression, logistic regression, random forest, and
support vector machine (SVM).
• Prediction of future cases: Use the rule to predict the output for
future inputs
• Knowledge extraction: The rule is easy to understand
• Compression: The rule is simpler than the data it explains
• Outlier detection: Exceptions that are not covered by the rule, e.g.,
fraud
 Unsupervised Learning
• Uses machine learning algorithms to analyze and cluster unlabeled
datasets (subsets called clusters).
• These algorithms discover hidden patterns or data groupings without the
need for human intervention.
• This method’s ability to discover similarities and differences in information
make it ideal for exploratory data analysis, cross-selling strategies,
customer segmentation, and image and pattern recognition.
• It’s also used to reduce the number of features in a model through the
process of dimensionality reduction.
• Principal component analysis (PCA) and singular value decomposition (SVD)
are two common approaches for this.
• Other algorithms used in unsupervised learning include neural networks, k-
means clustering, and probabilistic clustering methods.
 Unsupervised Learning
• Learning “what normally happens”
• No output
• Clustering: Grouping similar instances
• Example applications
• Customer segmentation in CRM
• Image compression: Color quantization
• Bioinformatics: Learning motifs
 Semi-Supervised Learning
• Semi-supervised learning offers a happy medium between supervised
and unsupervised learning.
• During training, it uses a smaller labeled data set to guide
classification and feature extraction from a larger, unlabeled data set.
• Semi-supervised learning can solve the problem of not having enough
labeled data for a supervised learning algorithm.
• It also helps if it’s too costly to label enough data.
 Reinforcement Machine Learning
• It is a machine learning model that is similar to supervised learning,
but the algorithm isn’t trained using sample data.
• This model learns as it goes by using trial and error.
• A sequence of successful outcomes will be reinforced to develop the
best recommendation or policy for a given problem.
• The IBM Watson® system that won the Jeopardy! challenge in 2011 is
a good example.
• The system used reinforcement learning to learn when to attempt an
answer (or question, as it were), which square to select on the board,
and how much to wager—especially on daily doubles.
 Reinforcement Machine Learning
• Reinforcement Learning
• Learning a policy: A sequence of outputs
• No supervised output but delayed reward
• Credit assignment problem
• Game playing
• Robot in a maze
• Multiple agents, partial observability, ...
 Differences ML vs DL vs NN
• Machine learning, deep learning, and neural networks are all sub-
fields of artificial intelligence.
• However, neural networks is actually a sub-field of machine learning,
and deep learning is a sub-field of neural networks.
• The way in which deep learning and machine learning differ is in how
each algorithm learns.
• "Deep" machine learning can use labeled datasets, also known as
supervised learning, to inform its algorithm, but it doesn’t necessarily
require a labeled dataset.
 Differences ML vs DL vs NN
• The deep learning process can ingest unstructured data in its raw
form (e.g., text or images), and it can automatically determine the set
of features which distinguish different categories of data from one
another.
• This eliminates some of the human intervention required and enables
the use of large amounts of data.
• You can think of deep learning as "scalable machine learning".
• Classical, or "non-deep," machine learning is more dependent on
human intervention to learn.
• Human experts determine the set of features to understand the
differences between data inputs, usually requiring more structured
data to learn.
 Differences ML vs DL vs NN
• Neural networks, or artificial neural networks (ANNs), are comprised
of node layers, containing an input layer, one or more hidden layers,
and an output layer.
• Each node, or artificial neuron, connects to another and has an
associated weight and threshold.
• If the output of any individual node is above the specified threshold
value, that node is activated, sending data to the next layer of the
network.
• Otherwise, no data is passed along to the next layer of the network
by that node.
• The “deep” in deep learning is just referring to the number of layers in
a neural network.
 Differences ML vs DL vs NN
• A neural network that consists of more than three layers—which
would be inclusive of the input and the output—can be considered a
deep learning algorithm or a deep neural network.
• A neural network that only has three layers is just a basic neural
network.
• Deep learning and neural networks are credited with accelerating
progress in areas such as computer vision, natural language
processing, and speech recognition.
• Deep learning and neural networks are credited with accelerating
progress in areas such as computer vision, natural language
processing, and speech recognition.
 Advantages and disadvantages of ML
• Depending on your budget, need for speed and precision required,
each algorithm type—supervised, unsupervised, semi-supervised, or
reinforcement—has its own advantages and disadvantages.
• For example, decision tree algorithms are used for both predicting
numerical values (regression problems) and classifying data into
categories.
• Decision trees use a branching sequence of linked decisions that may
be represented with a tree diagram.
• A prime advantage of decision trees is that they are easier to validate
and audit than a neural network.
• But they can be more unstable than other decision predictors.
 Advantages and disadvantages of ML
• Advantages
• These include ML identifying patterns and trends in massive volumes of data
that humans might not spot at all.
• And this analysis requires little human intervention: just feed in the dataset of
interest and let the machine learning system assemble and refine its own
algorithms—which will continually improve with more data input over time.
• Customers and users can enjoy a more personalized experience as the model
learns more with every experience with that person.
 Advantages and disadvantages of ML
• Disadvantages
• ML requires large training datasets that are accurate and unbiased.
• GIGO is the operative factor: garbage in / garbage out.
• Gathering sufficient data and having a system robust enough to run it might
also be a drain on resources.
• Machine learning can also be prone to error, depending on the input. With
too small a sample, the system could produce a perfectly logical algorithm
that is completely wrong or misleading.
• To avoid wasting budget or displeasing customers, organizations should act on
the answers only when there is high confidence in the output.
 Challenges of ML
• Technological singularity
• While not imminent, it raises ethical concerns, especially with autonomous
systems like self-driving cars.
• Accidents are inevitable—who bears responsibility?
• Should we pursue fully autonomous vehicles or limit them to assistive roles?
• The debate continues as AI evolves.

• AI impact on jobs
• AI may shift job demands rather than eliminate them.
• Like the shift from fuel to electric vehicles, AI will create new roles in
managing and improving AI systems.
• The challenge lies in helping workers transition to emerging job opportunities.
 Challenges of ML
• AI impact on jobs
• AI may shift job demands rather than eliminate them.
• Like the shift from fuel to electric vehicles, AI will create new roles in
managing and improving AI systems.
• The challenge lies in helping workers transition to emerging job opportunities.
• Privacy
• Privacy concerns have led to stronger data privacy and protection laws like
GDPR (2016) - was created to protect the personal data of people in the
European Union and European Economic Area and California Consumer
Privacy Act (CCPA) (2018)- requires businesses to inform consumers about
the collection of their data, giving individuals more control over their data.
• Businesses now prioritize security to prevent breaches, hacking, and
surveillance.
 Challenges of ML
• Bias and discrimination
• Ethical concerns: Bias in AI raises questions about fairness, as training data
often reflects human biases.
• Hiring bias: Amazon’s AI hiring tool unintentionally discriminated against
female candidates, leading to its discontinuation.
• Wider impact: Bias exists in various AI applications, including facial
recognition and social media algorithms.
• Corporate action: Companies like IBM are addressing AI ethics—IBM
discontinued general-purpose facial recognition to prevent misuse.
 Challenges of ML
• Accountability
• Lack of AI Regulations and Ethical Challenges
• No strict legislation: AI ethics lack enforcement due to the absence of
regulations.
• Corporate incentives: Companies follow ethical AI mainly to avoid negative
financial impacts.
• Ethical frameworks: Developed through collaboration but serve only as
guidelines.
• Challenges: Distributed responsibility and unforeseen consequences hinder
ethical AI implementation.
 Well Defined Learning Problems
• Definition: A computer program is said to learn from experience E
with respect to some class of tasks T and performance measure P, if
its performance at tasks in T, as measured by P, improves with
experience E.
• For example, a computer program that learns to play checkers might
improve its performance (ability to win) at the class of tasks involving
playing checkers games, through experience obtained by playing
games against itself.
• To have a well-defined learning problem, we must identity these
three features:
• The class of tasks
• The measure of performance to be improved
• The source of experience.
 Well Defined Learning Problems
• Examples

• A checkers learning problem:

• Task T: playing checkers
• Performance measure P: percent of games won against opponents
• Training experience E: playing practice games against itself
• A handwriting recognition learning problem:
• Task T: recognizing and classifying handwritten words within images
• Performance measure P: percent of words correctly classified
• Training experience E: a database of handwritten words with given classifications
• A robot driving learning problem:
• Task T: driving on public four-lane highways using vision sensors
• Performance measure P: average distance traveled before an error (as judged by human
overseer)
• Training experience E: a sequence of images and steering commands recorded while
observing a human driver
 Designing A Learning System
• Consider designing a program to learn to play checkers, with the goal of entering
it in the world checkers tournament
• Direct vs. Indirect Feedback
• Direct: Learns from labeled examples (e.g., checkers board states with correct moves).
• Indirect: Learns from outcomes, requiring credit assignment for past moves.
• Learner’s Control Over Training Data
• Teacher-led: Learner relies on expert-selected examples.
• Query-based: Learner asks for clarification on confusing cases.
• Self-play: Learner generates data by playing against itself.
• Training Data Distribution
• Learning is most effective when training data matches real-world test scenarios.
• Training only against itself may miss key situations encountered in real games.
• Final Decision
• The system will train by playing against itself, maximizing training data without needing an external
trainer.
 Designing A Learning System
• A checkers learning problem:
• Task T: playing checkers
• Performance measure P: percent of games won in the world
tournament
• Training experience E: games played against itself
• In order to complete the design of the learning system, we must now
choose
• The exact type of knowledge to be, learned
• A representation for this target knowledge
• A learning mechanism
 Designing A Learning System
• Choosing the Target Function
1. Learning Objective
1. The system must learn to choose the best move from legal moves.
2. Common in optimization problems like scheduling and process control.
2. Target Function Options
1. ChooseMove (B → M): Directly selects the best move (hard to learn).
2. TargetFunction V (B → ℝ): Assigns a numerical score to board states (easier to learn).
3. Defining V(b) for a Board State
1. Win: V(b) = 100
2. Loss: V(b) = -100
3. Draw: V(b) = 0
4. Non-final state: V(b) = V(b'), where b' is the best achievable final state assuming optimal play.
4. Challenges in Computing V
1. Requires searching all possible moves until the end of the game.
2. Not feasible for real-time decision-making → termed a nonoperational definition.
5. Function Approximation
1. The goal is to learn an operational form of V for practical use.
2. The learned function (denoted as ?) is an approximation of the ideal V.
 Designing A Learning System
• Choosing a Representation for the Target Function
• Choosing a Representation: Various options exist for representing the function c that
the learning program will learn.
• Possible Representations: A large table, rule-based system, quadratic polynomial, or
artificial neural network.
• Tradeoff: More expressive representations approximate V better but require more
training data.
• Decision: Use a simple representation—c as a linear combination of board features.
• x1: the number of black pieces on the board
• x2: the number of red pieces on the board
• x3: the number of black kings on the board
• x4: the number of red kings on the board
• x5: the number of black pieces threatened by red (i.e., which can be captured on red's next
turn)
• X6: the number of red pieces threatened by black
• Thus, our learning program will represent c(b) as a linear function of the form
 Designing A Learning System
• Choosing a Representation for the Target Function

• where w0 through W6 are numerical coefficients, or weights, to be chosen by the learning

algorithm.
• Learned values for the weights w1 through W6 will determine the relative importance of the
various board features in determining the value of the board, whereas the weight w0 will
provide an additive constant to the board value.
• Partial design of a checkers learning program:
• Task T: playing checkers
• Performance measure P: percent of games won in the world tournament
• Training experience E: games played against itself
• Target function: V: Board → R
• Target function representation
 Designing A Learning System
• Choosing a Function Approximation Algorithm
• To learn the target function f we require a set of training
examples, each describing a specific board state b and the
training value Vtrain(b) for b.
• Each training example is an ordered pair of the form <b,
Vtrain(b))>
• For instance, the following training example describes a
board state b in which black has won the game (note x2 =
0 indicates that red has no remaining pieces) and for
which the target function value Vtrain(b) is therefore +100.
 Designing A Learning System
• Estimating Training Values
• Training Information: Learner only knows if the game was won or lost.
• Challenge: Assigning scores to intermediate board states is unclear.
• Issue: A loss doesn’t mean all board states were bad; early moves could be strong.
• Solution: Estimate training value of a board state b as V(Successor(b)) using the current
approximation V.
• Justification: Works well if V is more accurate for states near the game's end.
• Convergence: Under certain conditions, this iterative method converges to perfect Vtrain
estimates.

• Rules for estimating training values.

• Vtrain(b) <- V(Successor(b))
 Designing A Learning System
• Adjusting weights
• Specify the learning algorithm for choosing the weights wi to best fit the set of
training examples {<b, Vtrain(b)>} as a first step (bestfit).
• One common approach is to define the best hypothesis, or set of weights, as
that which minimizes the squared error E between the training values and the
values predicted by the hypothesis V.

• There are many algorithms for adjusting the weights like LMS and gradient
descent etc.
Designing A Learning System
 Designing A Learning System
• Final Design of Checkers Learning System
1.Performance System
1. Solves the task (playing checkers) using the learned target function.
2. Takes a new game as input and produces game history as output.
3. Uses an evaluation function that improves performance over time.
2.Critic
1. Analyzes the game history and generates training examples.
2. Associates each game state with an estimated target function value.
3. Implements a training rule to refine the evaluation function.
3.Generalizer
1. Converts training examples into a generalized hypothesis.
2. Uses the LMS algorithm to learn the target function.
3. Produces an updated evaluation function based on learned weights.
4.Experiment Generator
1. Generates new problems (initial board states) for training.
2. Maximizes learning by selecting strategic positions.
3. Uses a simple or advanced strategy to improve learning efficiency.
 Designing A Learning System
• Design Choices & Constraints
• Uses a linear evaluation function with six board features.
• Limited by the expressiveness of the function representation.
• Capable of learning an approximation of the optimal function.
• Learning Potential
• Can effectively improve gameplay but unlikely to surpass human champions.
• More sophisticated representations (e.g., neural networks) enhance
performance.
• Similar methods applied to backgammon have led to competitive AI players.
 Designing A Learning System
• Final design of the checkers learning program.
 Designing A Learning System
• Sununary of choices in designing the checkers learning program.
 Issues in Machine Learning
• The field of machine learning, is concerned with answering
questions such as the following:
• What algorithms exist for learning general target functions from
specific training examples?
• In what settings will particular algorithms converge to the desired
function, given sufficient training data?
• Which algorithms perform best for which types of problems and
representations?
• How much training data is sufficient?
• What general bounds can be found to relate the confidence in
learned hypotheses to the amount of training experience and the
character of the learner's hypothesis space?
 Issues in Machine Learning
• When and how can prior knowledge held by the learner guide the
process of generalizing from examples?
• Can prior knowledge be helpful even when it is only approximately
correct?
• What is the best strategy for choosing a useful next training
experience, and how does the choice of this strategy alter the
complexity of the learning problem?
• What is the best way to reduce the learning task to one or more
function approximation problems? Put another way, what specific
functions should the system attempt to learn?
• Can this process itself be automated?
• How can the learner automatically alter its representation to
improve its ability to represent and learn the target function?
 History of Machine Learning
1.Early Foundations (1940s - 1950s)
1. 1943: McCulloch & Pitts propose the first artificial neuron model.
2. 1950: Alan Turing introduces the Turing Test for machine intelligence.
3. 1952: Arthur Samuel develops the first self-learning program (checkers-
playing AI).
2.Symbolic AI & Rule-Based Systems (1950s - 1970s)
1. 1957: Perceptron model introduced by Frank Rosenblatt (early neural
network).
2. 1960s: Early AI research focuses on rule-based learning and expert systems.
3. 1969: Minsky & Papert highlight perceptron limitations, slowing neural
network research.
 History of Machine Learning
3. Knowledge-Based Systems & Statistical Learning (1980s - 1990s)
1. 1980s: Development of decision trees and Bayesian networks.
2. 1986: Backpropagation algorithm popularized by Rumelhart, Hinton & Williams, reviving
neural networks.
3. 1990s: Introduction of Support Vector Machines (SVMs) and Random Forests.
4. Rise of Data-Driven Approaches (2000s - 2010s)
1. 2006: Geoffrey Hinton introduces Deep Learning, making neural networks viable again.
2. 2012: AlexNet wins ImageNet competition, marking a breakthrough in deep learning.
3. 2014: Generative Adversarial Networks (GANs) and Reinforcement Learning gain traction.
4. 2016: AlphaGo defeats human Go champion, showcasing deep reinforcement learning.
5. Modern Advances (2020s - Present)
1. Transformer models (GPT, BERT) revolutionize Natural Language Processing (NLP).
2. AI applications expand into healthcare, finance, autonomous systems, and robotics.
3. Ethical AI, interpretability, and bias mitigation become major research areas.
 History of Machine Learning
1. Early Foundations (1940s - 1950s)
1. 1943: McCulloch & Pitts propose the first artificial neuron model.
2. 1950: Alan Turing introduces the Turing Test for machine intelligence.
3. 1952: Arthur Samuel develops the first self-learning program (checkers-playing AI).
2. Symbolic AI & Rule-Based Systems (1950s - 1970s)
1. 1957: Perceptron model introduced by Frank Rosenblatt (early neural network).
2. 1960s: Early AI research focuses on rule-based learning and expert systems.
3. 1969: Minsky & Papert highlight perceptron limitations, slowing neural network research.
3. Knowledge-Based Systems & Statistical Learning (1980s - 1990s)
1. 1980s: Development of decision trees and Bayesian networks.
2. 1986: Backpropagation algorithm popularized by Rumelhart, Hinton & Williams, reviving neural networks.
3. 1990s: Introduction of Support Vector Machines (SVMs) and Random Forests.
4. Rise of Data-Driven Approaches (2000s - 2010s)
1. 2006: Geoffrey Hinton introduces Deep Learning, making neural networks viable again.
2. 2012: AlexNet wins ImageNet competition, marking a breakthrough in deep learning.
3. 2014: Generative Adversarial Networks (GANs) and Reinforcement Learning gain traction.
4. 2016: AlphaGo defeats human Go champion, showcasing deep reinforcement learning.
5. Modern Advances (2020s - Present)
1. Transformer models (GPT, BERT) revolutionize Natural Language Processing (NLP).
2. AI applications expand into healthcare, finance, autonomous systems, and robotics.
3. Ethical AI, interpretability, and bias mitigation become major research areas.
 History of Machine Learning
• 1. Early Foundations (1940s - 1950s)
• Theoretical Concepts & First Models
• 1943: Warren McCulloch & Walter Pitts introduce the first artificial
neuron model, laying the foundation for neural networks.
• 1950: Alan Turing proposes the Turing Test, a way to determine if a
machine exhibits intelligent behavior.
• 1951: Marvin Minsky & Dean Edmonds build the first neural network
computer (SNARC) using vacuum tubes.
• 1952: Arthur Samuel develops a self-learning checkers program, one
of the earliest examples of machine learning.
• 1957: Frank Rosenblatt invents the Perceptron, the first supervised
learning algorithm for pattern recognition.
 History of Machine Learning
• 2. Symbolic AI & Rule-Based Systems (1960s - 1970s)
• Emergence of Expert Systems & AI Winter
• 1960s:
• Research in AI focuses on symbolic reasoning and expert systems, using
manually encoded rules.
• First machine learning models based on decision trees appear.
• 1967: The nearest neighbor algorithm is introduced, enabling basic
pattern classification.
• 1969: Marvin Minsky & Seymour Papert publish Perceptrons, proving
that single-layer perceptrons are limited and cannot learn XOR
functions.
• This leads to decreased funding for neural networks, causing the first "AI
Winter" (1970s).
 History of Machine Learning
• 3. Knowledge-Based Systems & Statistical Learning (1980s - 1990s)
• Revival of Machine Learning & Probabilistic Models
• 1980s:
• Introduction of decision trees (e.g., ID3 algorithm by Ross Quinlan).
• Expert systems like MYCIN (medical diagnosis) gain popularity.
• Backpropagation algorithm (Hinton, Rumelhart, & Williams, 1986) allows multi-
layer neural networks to train effectively, reviving neural networks.
• 1990s:
• Shift from rule-based AI to statistical machine learning methods.
• Introduction of Support Vector Machines (SVMs) (Vladimir Vapnik, 1995).
• Development of Random Forests, an ensemble learning method.
• Naïve Bayes Classifier and Hidden Markov Models (HMMs) improve text and
speech recognition.
• Reinforcement Learning gains momentum, with Q-learning becoming a popular
technique.
 History of Machine Learning
• 4. Rise of Data-Driven Approaches (2000s - 2010s)
• Big Data & Deep Learning Revolution
• 2006: Geoffrey Hinton introduces Deep Learning, using Restricted Boltzmann
Machines (RBMs) for feature learning.
• 2009: The emergence of Big Data enables large-scale machine learning models.
• 2012: AlexNet (deep convolutional neural network) wins the ImageNet competition,
marking the breakthrough of Deep Learning in computer vision.
• 2013: Word2Vec, a deep learning technique for word embeddings, revolutionizes
Natural Language Processing (NLP).
• 2014:
• Ian Goodfellow introduces Generative Adversarial Networks (GANs) for synthetic data
generation.
• Deep Reinforcement Learning emerges, combining deep learning with decision-making
algorithms.
• 2015: Google’s AlphaGo defeats a professional human Go player, later defeating the
world champion in 2016.
• 2017: Google introduces the Transformer architecture, leading to breakthroughs in
NLP models like BERT and GPT.
 History of Machine Learning
• 5. Modern Advances (2020s - Present)
• AI in the Real World & Ethical Challenges
• 2020: GPT-3 is released, demonstrating human-like text generation with
175 billion parameters.
• 2021: AI models become more specialized in fields like medicine
(AlphaFold for protein folding) and autonomous vehicles.
• 2022:
• ChatGPT (GPT-3.5 & GPT-4) revolutionizes conversational AI.
• DALL·E 2 and Stable Diffusion improve AI-generated art.
• 2023-Present:
• AI ethics and regulations gain importance due to concerns over bias,
misinformation, and security.
• Multimodal AI (handling text, images, and video together) becomes the next
frontier.
• Self-supervised learning and neurosymbolic AI aim to improve explainability and
generalization in AI systems.
 Data Science vs Machine Learning
• Data Science and Machine Learning (ML) are closely related fields but
differ in their scope, techniques, and applications.
1. Definition
• Data Science: An interdisciplinary field that focuses on extracting
insights from structured and unstructured data using statistical
analysis, data engineering, and visualization.
• Machine Learning: A subset of artificial intelligence (AI) that enables
systems to learn patterns from data and make predictions or
decisions without explicit programming.
 Data Science vs Machine Learning
2. Scope & Focus
Aspect Data Science Machine Learning
Extract meaningful insights and solve Create predictive models that improve performance
Goal
business problems using data. over time.
Involves data collection, cleaning, Involves data collection, cleaning, visualization, and
Approach
visualization, and analysis. analysis.
Fields Statistics, Data Engineering, Business Deep Learning, Neural Networks, Supervised &
Covered Intelligence, Big Data, Machine Learning. Unsupervised Learning.
 Data Science vs Machine Learning
3. Techniques & Tools

Feature Data Science Machine Learning

Data wrangling, exploratory data Regression, classification,
Techniques analysis (EDA), hypothesis testing, data clustering, reinforcement
visualization. learning, neural networks.
Programming Languages Python, R, SQL, Julia. Python, R, MATLAB, C++.
Pandas, NumPy, Matplotlib, Tableau, Scikit-learn, TensorFlow,
Tools & Libraries
Power BI. PyTorch, Keras, OpenCV.
 Data Science vs Machine Learning
4. Key Differences
Aspect Data Science Machine Learning

Broader field involving data analysis and A specialized subset focused on

Nature
engineering. developing predictive models.
Works with structured, semi-structured, Works mainly with structured and
Data Handling
and unstructured data. labeled data.
Produces models that make
Generates insights, dashboards, and
Outcome predictions and automate decision-
reports.
making.
Advanced mathematics,
Mathematical Foundation Statistics, probability, linear algebra.
optimization, neural networks.
Business intelligence, fraud detection, Self-driving cars, speech recognition,
Use Cases
recommendation systems. anomaly detection.
 Data Science vs Machine Learning
5. Real-World Applications
Industry Data Science Application Machine Learning Application
Finance Risk analysis, credit scoring. Algorithmic trading, fraud detection.
Patient data analysis, medical Disease prediction, drug discovery.
Healthcare
research.
Customer segmentation, trend Product recommendations, price
E-commerce
analysis. optimization.
Sentiment analysis, customer Chatbots, targeted advertising.
Marketing
profiling.
 Data Science vs Machine Learning
6. Relationship Between Data Science & Machine Learning
• Machine Learning is a subset of Data Science.
• Data Science encompasses various processes like data preprocessing, feature
engineering, and interpretation, while ML focuses on model training and
predictions.
• Data Science uses Machine Learning models to derive insights and
make predictions, but ML also operates independently in AI-driven
applications.
• Not all Data Science projects use ML, but every ML project involves
Data Science techniques for data cleaning, preparation, and analysis.
 Data Science vs Machine Learning
• Choose Data Science if you are interested in data analysis,
visualization, and business insights.
• Choose Machine Learning if you want to develop AI models,
automate decision-making, and build predictive systems.
 Artificial Neural Networks (ANNs)
• Definition
• Artificial Neural Networks (ANNs) are computational models inspired by biological neural
networks. They consist of layers of interconnected nodes (neurons) that process and learn
from data.
• Theory
• ANNs are composed of an input layer, one or more hidden layers, and an output layer. Each
neuron applies a weighted sum of inputs followed by an activation function.
• Working Principles
1. Forward Propagation: Inputs pass through the network, undergoing weighted summation
and activation at each layer.
2. Backpropagation: Errors are propagated backward using optimization techniques like
gradient descent.
3. Training: The network adjusts weights based on the error to minimize loss.
• Applications
• Image and speech recognition
• Natural Language Processing (NLP)
• Autonomous vehicles
• Fraud detection
 Clustering
• Definition
• Clustering is an unsupervised machine learning technique that groups data points based on
similarity.
• Theory
• Clustering algorithms identify inherent patterns in data and form clusters based on distance
metrics.
• Working Principles
1. Partitioning: Data points are assigned to clusters using algorithms like k-means, hierarchical
clustering, or DBSCAN.
2. Distance Calculation: Measures like Euclidean or cosine similarity determine cluster
membership.
3. Cluster Refinement: Re-evaluation ensures optimal clustering.
• Applications
• Customer segmentation
• Anomaly detection
• Market research
• Image segmentation
 Reinforcement Learning (RL)
• Definition
• Reinforcement Learning (RL) is a machine learning paradigm where an agent learns optimal
actions through trial and error in an environment.
• Theory
• RL is modeled as a Markov Decision Process (MDP), consisting of states, actions, rewards, and
policies.
• Working Principles
1. Agent-Environment Interaction: The agent takes actions in an environment and receives
rewards.
2. Policy Optimization: The agent refines its strategy to maximize cumulative rewards.
3. Exploration vs. Exploitation: The agent balances discovering new actions and using known
best actions.
• Applications
• Robotics
• Game AI (e.g., AlphaGo, OpenAI Gym)
• Autonomous vehicles
• Financial trading
 Decision Tree Learning
• Definition
• A Decision Tree is a supervised learning model used for classification and regression
tasks.
• Theory
• Decision Trees split data recursively based on feature values to form a tree structure.
• Working Principles
1.Node Splitting: The best attribute is chosen using measures like Gini Index or
Entropy.
2.Recursive Partitioning: The dataset is divided into smaller subsets.
3.Pruning: Reduces overfitting by removing unnecessary branches.
• Applications
• Medical diagnosis
• Credit scoring
• Spam filtering
• Sentiment analysis
 Bayesian Networks
• Definition
• Bayesian Networks (BNs) are probabilistic graphical models that represent
dependencies among variables using directed acyclic graphs.
• Theory
• Each node represents a variable, and directed edges signify conditional
dependencies governed by Bayes' Theorem.
• Working Principles
1.Probability Distribution: Captures joint probability distributions.
2.Conditional Independence: Uses directed edges to model dependencies.
3.Inference: Probabilistic reasoning determines likely outcomes.
• Applications
• Medical diagnosis
• Risk assessment
• Speech recognition
• Fraud detection
 Support Vector Machines (SVMs)
• Definition
• Support Vector Machines (SVMs) are supervised learning models used for classification and
regression by finding the optimal decision boundary.
• Theory
• SVMs maximize the margin between classes using a hyperplane in high-dimensional space.
• Working Principles
1. Linear SVM: Uses a straight hyperplane for classification.
2. Non-Linear SVM: Uses the kernel trick (e.g., RBF, polynomial) to transform data into higher
dimensions.
3. Support Vectors: Data points closest to the decision boundary that influence classification.
• Applications
• Face detection
• Text categorization
• Handwriting recognition
• Bioinformatics
 Genetic Algorithms (GAs)
• Definition
• Genetic Algorithms (GAs) are optimization techniques inspired by natural selection and
evolution.
• Theory
• GAs evolve solutions using genetic operators like selection, crossover, and mutation.
• Working Principles
1. Initialization: A population of potential solutions is generated.
2. Selection: The best individuals are chosen based on fitness scores.
3. Crossover & Mutation: New solutions are generated through recombination and random
mutations.
4. Convergence: The algorithm iterates until an optimal solution is found.
• Applications
• Scheduling problems
• Feature selection
• Robotics and AI
• Game development
 Generative AI
• Generative AI has capabilities, the system have:
• A content generator that can generate text, images and other
content based on the data it was trained on?
• Automated classification to read and classify written input, such as
evaluating and sorting customer complaints or reviewing customer
feedback sentiment?
• A summary generator that can transform dense text into a high-
quality summary, capture key points from financial reports, and
generate meeting transcriptions?
• A data extraction capability to sort through complex details and
quickly pull the necessary information from large documents?
• Common machine learning algorithms

• A number of machine learning algorithms are commonly used. These include:

• Neural networks simulate the way the human brain works, with a huge number of linked processing nodes. Neural networks are good at recognizing patterns and play an important role in applications including natural language translation, image recognition, speech recognition,
and image creation.

• Linear regression

• This algorithm is used to predict numerical values, based on a linear relationship between different values. For example, the technique could be used to predict house prices based on historical data for the area.

• Logistic regression

• This supervised learning algorithm makes predictions for categorical response variables, such as “yes/no” answers to questions. It can be used for applications such as classifying spam and quality control on a production line.

• Clustering

• Using unsupervised learning, clustering algorithms can identify patterns in data so that it can be grouped. Computers can help data scientists by identifying differences between data items that humans have overlooked.

• Decision trees

• Decision trees can be used for both predicting numerical values (regression) and classifying data into categories. Decision trees use a branching sequence of linked decisions that can be represented with a tree diagram. One of the advantages of decision trees is that they are easy
to validate and audit, unlike the black box of the neural network.

• Random forests

• In a random forest, the machine learning algorithm predicts a value or category by combining the results from a number of decision trees.

• Real-world machine learning use cases

• Here are just a few examples of machine learning you might encounter every day:

• Speech recognition: It is also known as automatic speech recognition (ASR), computer speech recognition, or speech-to-text, and it is a capability which uses natural language processing (NLP) to translate human speech into a written format. Many mobile devices incorporate
speech recognition into their systems to conduct voice search—e.g. Siri—or improve accessibility for texting.

• Customer service:Online chatbots are replacing human agents along the customer journey, changing the way we think about customer engagement across websites and social media platforms. Chatbots answer frequently asked questions (FAQs) about topics such as shipping, or
provide personalized advice, cross-selling products or suggesting sizes for users. Examples include virtual agents on e-commerce sites; messaging bots, using Slack and Facebook Messenger; and tasks usually done by virtual assistants and voice assistants.

• Computer vision: This AI technology enables computers to derive meaningful information from digital images, videos, and other visual inputs, and then take the appropriate action. Powered by convolutional neural networks, computer vision has applications in photo tagging on
social media, radiology imaging in healthcare, and self-driving cars in the automotive industry.

• Recommendation engines: Using past consumption behavior data, AI algorithms can help to discover data trends that can be used to develop more effective cross-selling strategies. Recommendation engines are used by online retailers to make relevant product
recommendations to customers during the checkout process.

• Robotic process automation (RPA): Also known as software robotics, RPA uses intelligent automation technologies to perform repetitive manual tasks.

• Automated stock trading: Designed to optimize stock portfolios, AI-driven high-frequency trading platforms make thousands or even millions of trades per day without human intervention.

• Fraud detection: Banks and other financial institutions can use machine learning to spot suspicious transactions. Supervised learning can train a model using information about known fraudulent transactions. Anomaly detection can identify transactions that look atypical and
deserve further investigation.

• How to choose the right AI platform for machine learning

• Selecting a platform can be a challenging process, as the wrong system can drive up costs, or limit the use of other valuable tools or technologies. When reviewing multiple vendors to select an AI platform, there is often a tendency to think that more features = a better system.
Maybe so, but reviewers should start by thinking through what the AI platform will be doing for their organization. What machine learning capabilities need to be delivered and what features are important to accomplish them? One missing feature might doom the usefulness of
an entire system. Here are some features to consider.