LEARNING

CONTENT

1. Learning
2. Forms of Learning
3. Theory of Learning
4. PAC Learning
5. Introduction to Statistical Learning
6. Introduction to Reinforcement
Learning
7. Learning from Rewards
8. Passive Reinforcement Learning
9. Active Reinforcement Learning
RECAP
• Planning
• Planning with State Space Search
• Partial Order Planning
• Hierarchical Planning
• Conditional Planning
LEARNING
• Definition of Learning: The process of acquiring
knowledge or skills through experience, study,
or instruction.
• Importance of Learning in AI: Learning enables
AI systems to adapt and improve their
performance based on data and feedback.
INTRODUCTION TO LEARNING
• Learning is a fundamental process in human
cognition and development.
• Learning encompasses the acquisition of
knowledge, skills, behaviors, and attitudes
through experience, study, or instruction.
• It plays a crucial role in personal growth,
education, and professional advancement.
• This presentation will delve into different
forms and theories of learning to provide a
comprehensive understanding of this complex
phenomenon.
FORMS OF LEARNING
Classical
Conditioning
1. Classical
Conditioning
2. Operant Conditioning Cognitive FORMS OF Operant
3. Observational Learning LEARNING Conditioning

Learning
4. Cognitive Learning
Observational
Learning
FORMS OF LEARNING
• Classical Conditioning
• Classical conditioning involves learning by association, where a neutral stimulus
becomes associated with a meaningful stimulus to elicit a response.
• Example: A dog learns to salivate at the sound of a bell because it has been associated
with the presentation of food.

• Operant Conditioning
• Operant conditioning focuses on learning through consequences, where behaviors are
strengthened or weakened based on their consequences.
• Example: A child learns to clean up their toys after playing to receive praise or avoids
touching a hot stove after getting burned.
FORMS OF LEARNING
• Observational Learning
• Observational learning occurs when individuals learn by observing others' behaviors
and their consequences.
• Example: Children learn to tie their shoes by watching their parents or peers
demonstrate the process.

• Cognitive Learning
• Cognitive learning involves mental processes like thinking, reasoning, and problem-
solving.
• Example: Students learn math concepts by understanding the underlying principles and
applying them to solve problems.
GENERAL MODEL OF LEARNING AGENTS
Learning Element:
Adds knowledge, makes improvement to system
Performance Element:
Performs task, selects external actions
Critic:
Monitors results of performance, provides feedback to learning element
Problem Generator:
Actively suggests experiments, generates examples to test
Performance Standard:
Method / standard of measuring performance
INDUCTIVE LEARNING
• Inductive learning takes the traditional sequence of
a lesson and reverses things. Instead of saying,
“Here is the knowledge; now go practice it,”
inductive learning says, “Here are some objects,
some data, some artifacts, some experiences… what
knowledge can we gain from them?”
• Seeing several red apples and concluding that all
apples are red.
• Example: Predicting the weather based on past
observations.
LEARNING DECISION TREES
• Decision trees help make decisions by mapping
out choices based on features.
• A decision tree takes as input an object or
situation described by a set of properties, and
outputs a yes/no "decision.
• Decision trees therefore represent Boolean
functions. Functions with a larger range of
outputs can also be represented, but for
simplicity we will usually stick to the Boolean
case.
USING INFORMATION THEORY IN
LEARNING
• Information theory, also known as the mathematical
theory of communication, is an approach that
studies data processing and measurement in the
transmission of information.
• Information theory helps measure and analyze the
amount of information in data.
• It guides us in selecting the most important features
or attributes for learning.
• Example: Identifying keywords in emails to classify
them as spam or non-spam.
LEARNING GENERAL LOGIC
DESCRIPTIONS
• Learning general logic descriptions involves inferring logical rules or patterns
from data.
• These rules help in making decisions or predictions based on observed examples.
• Example: Creating rules for a chatbot to respond to customer queries based on past
interactions.
THEORY OF LEARNING
• Behaviorism
• Focuses on observable behaviors and the impact of external stimuli on learning.
• Example: Teaching a dog to sit by rewarding it with a treat when it follows the command.
• Cognitivism
• Emphasizes mental processes like memory, attention, and problem-solving in learning.
• Example: Memorizing multiplication tables by practicing and repeating them.
• Constructivism
• Views learning as an active process of constructing knowledge through experiences and
interactions.
• Example: Building a model airplane from a kit by following instructions and trial-and-
error.
PAC LEARNING
• Probably Approximately Correct (PAC) learning is a theoretical framework
for studying the computational complexity of learning algorithms.
• It aims to find hypotheses that are probably approximately correct based on a
limited sample of data, with high probability.
• Example: Predicting whether an email is spam or not based on a sample of
labeled emails.
INTRODUCTION TO STATISTICAL
LEARNING
• Statistical learning involves learning from
data to make predictions or decisions.
• It includes techniques like supervised
learning (predicting outcomes), unsupervised
learning (finding patterns), and semi-
supervised learning (mixing labeled and
unlabeled data).
• Example: Predicting house prices based on
historical sales data and property features.
INTRODUCTION TO REINFORCEMENT
LEARNING
• Reinforcement learning is learning through trial
and error by interacting with an environment to
achieve goals.
• It involves taking actions, receiving feedback in
the form of rewards or penalties, and adjusting
behavior to maximize cumulative rewards over
time.
• Example: Teaching a robot to navigate through a
maze by rewarding it with points for reaching the
goal.
LEARNING FROM REWARDS
• Learning from rewards involves adjusting
behavior based on the consequences received.
• Positive rewards reinforce desired actions,
while negative rewards discourage unwanted
behavior.
• Example: A dog learning tricks such as sitting
or rolling over in exchange for treats.
PASSIVE REINFORCEMENT LEARNING
• Passive reinforcement learning involves learning from the environment without taking actions.
• It focuses on estimating the value of different states or actions based on observed data.
• Example: Estimating the value of different chess board positions without making any moves.
ACTIVE REINFORCEMENT LEARNING
• Active reinforcement learning requires taking actions to maximize cumulative rewards.
• It involves a trade-off between exploring new actions and exploiting known actions.
• Example: Learning to play a video game by trying different strategies and adapting based
on the game's feedback.
LEARNING IN NEURAL NETWORKS
• Neural networks are computational models inspired by the structure and
function of the human brain.
• These networks consist of interconnected nodes (neurons) organized in
layers, which process information through weighted connections.
• Example: Teaching a neural network to classify images by exposing it to
labeled examples of different objects.
LEARNING IN BELIEF NETWORKS
• Belief networks, also known as Bayesian
networks, model probabilistic relationships
between variables.
• These networks use probabilistic inference to
represent uncertain knowledge and make
predictions or decisions.
• Example: Predicting the likelihood of a disease
based on symptoms, medical history, and
demographic data using a Bayesian network.
KNOWLEDGE IN LEARNING
• Prior knowledge plays a vital role in guiding and shaping the learning
process.
• Leveraging existing knowledge allows learners to build upon established
foundations, facilitating deeper understanding and faster acquisition of new
information.
• Example: Applying prior knowledge of basic arithmetic to learn more
advanced mathematical concepts such as calculus.
SIMPLE EXAMPLES OF LEARNING

• Learning to play a musical instrument by practicing scales and melodies,

gradually improving proficiency over time.
• Learning to cook by following recipes and experimenting with different
ingredients and techniques.
• Learning to drive a car by attending driving lessons and gaining experience
behind the wheel.
EXPLANATION-BASED LEARNING
• Explanation-based learning involves deriving general rules or principles from
specific instances or examples.
• By understanding the underlying reasons or principles behind observed
phenomena, learners can generalize and apply their knowledge to new
situations.
• Example: Learning to solve complex mathematical problems by
understanding the underlying mathematical principles and applying them
systematically.
LEARNING USING RELEVANCE
INFORMATION
• Relevance information helps identify significant features or patterns in data
that are crucial for learning and decision-making.
• By focusing on relevant information, learners can streamline their learning
process, improve efficiency, and achieve better outcomes.
• Example: Identifying key market trends and customer preferences to develop
targeted marketing strategies for specific demographic groups.
INDUCTIVE LOGIC PROGRAMMING
• Inductive logic programming combines
elements of logic programming and machine
learning to induce logical rules from examples.
• This approach enables the automatic generation
of logical rules from observed data, facilitating
automated reasoning and decision-making.
• Example: Automatically deriving rules for
diagnosing medical conditions based on patient
symptoms and medical records using inductive
logic programming.
REAL-LIFE APPLICATIONS OF LEARNING
• Learning has numerous applications in various domains: Education: Adaptive
learning platforms personalize learning experiences based on individual
progress and preferences.
• Healthcare: Predictive analytics assist in diagnosing diseases and
recommending personalized treatment plans.
• Finance: Machine learning algorithms analyze financial data to detect
fraudulent transactions and predict market trends.
FUTURE DIRECTIONS IN LEARNING
• Lifelong Learning: Emphasizing continuous learning throughout life enables
individuals to adapt to evolving technologies and societal changes.
• Interdisciplinary Research: Collaboration between experts in various fields
fosters innovative solutions to complex learning problems.
• Ethical AI: Developing ethical frameworks and regulations ensures that AI
technologies benefit society while minimizing harm and bias.
CONCLUSION
• Learning is a multifaceted journey that shapes human
development and technological progress.
• Let's harness our understanding of learning to tackle
challenges, innovate, and create a better future.
• Thank you for joining us on this enlightening
exploration. Let's continue to learn, adapt, and thrive
together.
EDUCATE

EMPOWER

ILLUMINATE