INTRODUCTION TO

ARTIFICIAL INTELLIGENCE

Mahesh Madhushan
B.Sc. in ICT (RUSL), B.Eng. (Hons) in Software Engineering (UK), M.Sc. in IT (UOP)
 INTRODUCTION TO
ARTIFICIAL INTELLIGENCE

Session 02 – AI Techniques
 Overview of AI
Techniques
 AI techniques refer to the
methods and approaches used
to enable machines to perform
intelligent tasks.
 Importance:
 These techniques empower AI
systems to learn, understand
language, perceive visual
information, and interact with
the physical world.
Overview of AI Techniques

 AI techniques are typically divided into two

categories:
 Symbolic AI: Symbolic AI techniques use symbols to
represent knowledge and reasoning.
 Subsymbolic AI: Subsymbolic AI techniques use
numerical representations to represent knowledge and
reasoning.
Symbolic AI Techniques

Rule-based systems: Rule-based systems use a set

of rules to represent knowledge and reasoning.

Expert systems: Expert systems are computer

programs that use a set of rules to solve problems
in a specific domain of expertise.

Logic-based systems: Logic-based systems use

logical reasoning to solve problems.
Subsymbolic AI Techniques

Neural networks: Neural networks are inspired by the

human brain. They are composed of a large number of
interconnected nodes, which can learn from experience.

Fuzzy logic: Fuzzy logic is a type of logic that allows for

partial truth values. This makes it useful for representing
and reasoning about uncertain or imprecise information.

Evolutionary computation: Evolutionary computation is a

type of optimization algorithm that is inspired by the
process of natural selection.
Hybrid AI Techniques

Hybrid AI techniques combine symbolic and

subsymbolic AI techniques.

This can be done to improve the performance of

AI systems.

For example, a hybrid AI system might use a

neural network to learn from experience and a
rule-based system to provide logical reasoning.
Types of AI Techniques
 Machine Learning
 Supervised Learning
 Unsupervised Learning
 Reinforcement Learning
 Deep Learning
 Expert Systems
 Computer Vision
 Natural Language Processing
(NLP)
 Evolutionary Algorithms
 Each type's role in solving
different problems
 Machine Learning
 Machine Learning is a subset of artificial intelligence that
focuses on the development of algorithms that allow
computers to learn from and make predictions or decisions
based on data.
 Key Components:
 Data: Machine learning algorithms require large amounts of data
to learn patterns and make accurate predictions.
 Algorithms: Various machine learning algorithms, such as
supervised learning, unsupervised learning, and reinforcement
learning, are used to train models.
 Model Evaluation: The performance of machine learning models
is assessed using metrics such as accuracy, precision, recall, and
F1 score.
 How Does Machine Learning Work?
 Machine learning works by training models on data.
 Raw data is fed into the model, which identifies patterns
and relationships.
 Themodel is then tested on new data to evaluate its
accuracy.
 Basedon the results, the model can be refined and
improved.
 Types of Machine
Learning
 Supervised Learning :
 In supervised learning, the model is
trained on labeled data to train models
for prediction, where the input features
are associated with corresponding target
labels. Common algorithms include
linear regression, logistic regression,
decision trees, and support vector
machines.
 Input: Labeled data (e.g., (image, "cat"))
 Algorithm: Learns a mapping from inputs to
outputs (e.g., a classification model for cat
images)
 Examples:
 Image classification
 Spam detection
Types of Machine Learning
 Types of Machine
Learning
 Unsupervised Learning :
 Training models on unlabeled data to discover hidden
patterns or structures. Clustering algorithms (e.g., K-
means clustering, hierarchical clustering) and
dimensionality reduction techniques (e.g., Principal
Component Analysis) are examples of unsupervised
learning.
 Input: Unlabeled data (e.g., a collection of images)
 Algorithm: Discovers hidden patterns and structures in the
data (e.g., clustering algorithms for finding groups of
similar images)
 Examples:
 Clustering
 Dimensionality reduction
 Anomaly detection
Types of Machine Learning
 Types of Machine Learning
 Reinforcement Learning
 Reinforcement learning is a type of machine learning where an agent learns to
make decisions by interacting with an environment. The agent receives
feedback in the form of rewards or penalties based on its actions, and its goal is
to maximize cumulative reward over time.

 Input: Environment (e.g., a game or robot simulation)

 Algorithm: Learns through trial and error, maximizing a reward function (e.g.,
winning games, navigating efficiently)
 Examples:
 Game playing (e.g., AlphaGo)
 Robot control
 Financial trading
Types of Machine Learning
Machine Learning - Real-
World Applications
 From personalized recommendations to disease
prediction, machine learning is transforming
industries.
 Healthcare: Analyzing medical images, predicting
patient risks, and even assisting in surgery.
 Finance: Detecting fraud, making smarter
investment decisions, and personalizing financial
advice.
 Retail: Recommending products, optimizing
inventory management, and enhancing customer
experience.
 Deep Learning
 A subfield of machine learning inspired by the structure and
function of the human brain.
 Deep learning models use artificial neural networks with
multiple layers (hence "deep") to learn complex patterns and
relationships in data
 Deep learning has transformed fields like computer vision,
natural language processing, and speech recognition, and its
applications continue to grow rapidly
 Deep learning models rely on artificial neural networks
(ANNs).
 Deep Learning - Key Components
 Neural Networks: Deep learning models are composed of
interconnected layers of artificial neurons that process input data
and extract hierarchical features.
 Deep Learning Architecture: Deep neural networks can have many
hidden layers, allowing them to learn complex patterns and
representations from raw data.
 Representation Learning: Deep learning algorithms automatically
learn feature representations from data, eliminating the need for
manual feature engineering.
 Training Process: Deep learning models are trained using large
datasets through the process of backpropagation, where errors are
propagated backward through the network to update the model's
parameters.
 Convolutional neural networks
(CNNs): Used for image and
video analysis

Deep Recurrent neural networks

Learning (RNNs): Used for sequential
data, such as text and speech
Architectures
Transformers: Used for natural
language processing and
computer vision
 Training Deep Learning Models: Mimicking the
Brain's Learning Process

Deep learning models learn through a process

similar to how the human brain learns.
The
Learning Large amounts of data are fed into the network.

Process The network adjusts the weights of connections

between artificial neurons to improve accuracy.

The process is iterative, with the model

continuously learning and improving.
The Learning Process
 Deep Learning - Application
 Image recognition
 Natural language processing (NLP)
 Speech recognition
 Machine translation
 Recommendation systems
 Autonomous Systems
 self-driving cars, drones, and robots
 Healthcare
 disease diagnosis
 medical image analysis
 drug discovery
 personalized treatment recommendation
 Artificial Neural Networks

 Artificial
Neural Networks (ANNs) are computational
models inspired by the structure and function of the
human brain, composed of interconnected nodes
(neurons) organized in layers.

 Key Concepts:
 Neurons: Basic processing units
 Layers: Groups of neurons
 Activation functions: Determine neuron output
 Backpropagation: Training algorithm
Artificial Neural Networks
Artificial Neural Networks
Types of Artificial Neural Networks
 Feedforward Neural Networks: Feedforward neural networks are the simplest type
of ANN, where information flows in one direction from input to output layers
without cycles or feedback loops. They are used for tasks such as classification,
regression, and pattern recognition.
 Recurrent Neural Networks (RNNs): Recurrent neural networks have connections
that form directed cycles, allowing them to process sequential data such as time
series, natural language, and speech. They are used in tasks such as language
modeling, machine translation, and speech recognition.
 Convolutional Neural Networks (CNNs): Convolutional neural networks are
specialized for processing grid-like data such as images, consisting of convolutional
layers followed by pooling layers. They are used for tasks such as image
classification, object detection, and image segmentation.
 Long Short-Term Memory (LSTM) Networks: LSTM networks are a type of recurrent
neural network with specialized memory cells that can retain information over long
sequences, making them effective for tasks such as sequence prediction, language
modeling, and time series forecasting.
 Natural Language Processing (NLP)

 NLP is a subfield of AI that deals with the understanding, interpretation, and

generation of human language.

 Core Tasks of NLP

 Text Preprocessing: Cleaning and preparing text data for analysis.
 Text Analysis: Breaking down text into its components (words, sentences).
 Feature Engineering: Extracting meaningful features from text data.
 Model Training: Training NLP models on large amounts of data.
 Evaluation: Assessing the performance of NLP models.

 Applications: Chatbots, Sentiment Analysis, Language Translation, Summarization,

Spam detection
 Computer Vision
 Computer Vision is a field of artificial intelligence that
enables computers to interpret and understand visual
information (images and videos) from the real world.
 Core Components:
 Image Processing
 Object Detection
 Image Segmentation
 Image Classification

 Applications: Facial Recognition, Autonomous Vehicles,

Medical Image Analysis
Robotics
 Robotics is a subfield of AI that deals with the design,
construction, operation, and application of robots to perform
tasks in various domains.
 Components of a robot:
 Mechanical structure
 Sensors
 Actuators
 Control system
 Applications: Industrial Automation, Healthcare, Space
Exploration
 Robot Types
 Industrial Robots: Workhorses of automation, performing
repetitive tasks in manufacturing. (Think: car assembly lines)
 Service Robots: Assist humans in various tasks, from domestic
chores (vacuuming robots) to customer service (chatbots).
 Surgical Robots: Enhance precision and minimally invasive
techniques in surgery.
 Humanoid Robots: Mimic human form and movement, still
under development but with promising potential.
 Mobile robots: Can move autonomously, such as self-driving
cars and drones
 Expert Systems
 An expert system is a computer program that simulates the
knowledge and reasoning abilities of a human expert in a specific
domain.
 The Building Blocks of an Expert System
 Knowledge base: Contains facts and rules about the domain
 Inference engine: Uses the knowledge base to reason and make
decisions
 User Interface: Enables users to interact with the system

 Applications: Medical Diagnosis, Technical Support, Financial

Planning, Equipment Maintenance, Customer Service, Legal
reasoning, Scientific research
Challenges and Considerations

 While powerful, AI techniques raise ethical and practical

considerations that demand our attention.
 Bias and fairness: Ensuring algorithms are unbiased and
treat everyone fairly.
 Explainabilityand interpretability: Understanding how AI
models arrive at their decisions.
 Privacyand security: Protecting sensitive data and
preventing misuse of AI systems.
Questions?

Thank You