Scribd
Upload
4 views

Machine learning Lecture

Uploaded by

Asma Ayub
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
4 views

Machine learning Lecture

Uploaded by

Asma Ayub
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 20

Introduction to Machine Learning

Dr.Sadaqat Ali Ramay


What is Machine
Learning?

● Branch of artificial intelligence


● Allows systems to learn and improve from
experience
● Uses data to build predictive models
● No explicit programming required
● What types of machine learning have you
heard of before?
Types of Machine Learning

● Supervised Learning
● Unsupervised Learning
● Reinforcement Learning
● Which type do you think is used to teach computers to play
video games?
Supervised Learning

● Uses labeled data


● Goal: Learn mapping from inputs to
outputs
● Examples: Regression, classification
● Applications: Spam detection, price
prediction
● Can you think of other real-world examples
of supervised learning?
Unsupervised Learning

● No labeled data
● Goal: Find patterns or structures in data
● Examples: Clustering, dimensionality reduction
● Applications: Customer segmentation, anomaly detection
● How might businesses use unsupervised learning?
Reinforcement Learning

● Learn through interaction with environment


● Goal: Maximize reward
● Examples: Markov decision processes, Q-
learning
● Applications: Robotics, game-playing AI
● What games do you think AI could learn to
play using this method?
What is Statistical Pattern Recognition?

● Field focused on classifying and analyzing patterns


● Uses statistical techniques
● Rooted in statistical theory and probability
● How do you think this relates to machine learning?
Steps in Statistical Pattern
Recognition

● 1. Data Representation
● 2. Feature Extraction and Selection
● 3. Model Building
● 4. Validation
● Which step do you think is most important?
Why?
Connections Between ML and SPR

● Both rely on statistical principles


● Share common models (e.g. regression, decision trees)
● Similar applications (image recognition, NLP)
● Can you see how these fields might work together?
Applications of ML and
Pattern Recognition

● Computer Vision
● Natural Language Processing
● Bioinformatics
● Finance
● Healthcare
● Which application interests you most?
What are Graphical Models?

● Probabilistic models using graphs


● Combine graph theory and probability
● Model relationships among variables
● Nodes represent variables
● Edges represent dependencies
● How might graphs help visualize complex relationships?
Types of Graphical
Models

● Directed Graphical Models (Bayesian


Networks)
● Undirected Graphical Models (Markov
Random Fields)
● What do you think the difference between
directed and undirected means?
Bayesian Networks (Full Bayes)

● Directed acyclic graph (DAG)


● Nodes are random variables
● Edges show direct dependencies
● Use conditional probabilities
● Example: Disease diagnosis model
● How might a doctor use this type of model?
Advantages of Bayesian Networks

● Handle uncertainty explicitly


● Support two-way reasoning (causal and diagnostic)
● Efficiently represent complex relationships
● Can you think of a situation where handling uncertainty is
important?
What is Naïve Bayes?

● Simplified Bayesian model


● Assumes features are conditionally
independent
● Fast and simple to compute
● Works well in many real-world scenarios
● Why do you think it's called "naïve"?
Naïve Bayes Applications

● Spam filtering
● Sentiment analysis
● Text classification
● Have you ever wondered how your email filters out spam?
Comparing Full Bayes and Naïve Bayes

● Full Bayes:
● More complex
● Models all dependencies
● Can be more accurate with enough data
● Naïve Bayes:
● Simpler
● Assumes independence
● Works well with limited data
● Which do you think is used more often in real applications?
Advantages of Graphical Models

● Provide visual representation of relationships


● Simplify complex probability calculations
● Allow for intuitive understanding of systems
● Support various inference tasks
● How might these advantages help in problem-solving?
Challenges in Machine
Learning

● Collecting and preparing large datasets


● Choosing appropriate models and features
● Avoiding overfitting
● Interpreting complex models
● What challenges do you foresee in
developing AI systems?
The Future of ML and Pattern Recognition

● Continued advancements in AI capabilities


● Integration with other technologies (IoT, big data)
● Ethical considerations and responsible AI
● Potential for solving complex global problems
● How do you think these fields will impact your future?

You might also like