Introduction To Machine Learning
Member, AI in Education
24th January 2025
OUTLINE 1. Introduction
2. Machine Learning & Its Types
3. Data processing
4. Popular machine learning frameworks
5. Ethics in machine learning
6. Future of machine learning
7. Challenges in machine learning
8. Conclusion
What is Machine Learning?
Machine learning is a subset of artificial
intelligence that enables systems to learn and
improve from experience without being
explicitly programmed.
It focuses on building algorithms that can
process data, identify patterns, and make
decisions.
3
History of Machine Learning
4
Types of Machine Learning
7
Supervised Learning
Supervised learning is a type of machine learning
where the model is trained on labeled data. Each
input is paired with a corresponding output
(label), and the model learns to map inputs to
outputs.
Common Algorithms
• Linear Regression
• Logistic Regression
• Decision Trees
• Support Vector Machines (SVM)
• Random Forests
• Neural Networks
8
Unsupervised Learning
This is a type of machine learning where the
model works with unlabeled data. It identifies
hidden patterns, structures, or groupings within
the data without predefined outputs.
Practical Examples:
• Customer Segmentation: Businesses use
clustering algorithms to segment customers
based on purchasing behavior, helping to
tailor marketing strategies.
• Image Compression: Used to reduce the size
of image files while maintaining quality.
• Recommender Systems: Unsupervised
learning can help in identifying user
preferences based on item similarities, leading
to better product recommendations.
9
Reinforcement Learning
Reinforcement learning is a type of
machine learning where an agent
learns by interacting with its
environment. The agent takes
actions and receives feedback in the
form of rewards or penalties,
optimizing its behavior over time.
10
Reinforcement Learning Applications
Personalized Learning: RL can be used to tailor
educational content to individual student needs.
By analyzing student interactions, the system can
adapt the learning path to optimize engagement
and retention.
Intelligent Tutoring Systems: These systems can
provide real-time feedback and adjust difficulty
levels based on student performance, helping
them learn more effectively.
Gamified Learning: Implementing RL principles in
educational games can motivate students by
providing rewards for progress, encouraging
them to explore and learn at their own pace.
Curriculum Development: By analyzing how
students interact with different materials,
educational institutions can adapt curricula to
focus on topics that yield the best learning
outcomes.
Application of ML in Finance
12
Application of ML in Agriculture
AI for Irrigation Plant Disease Detection
13
Application of ML in Healthcare
14
Key Concepts of Machine Learning
Features
Algorithm
Individual measurable properties or
The method or procedure used to train the
characteristics of the data used as input (e.g.,
model (e.g., linear regression, decision trees).
age, salary, or pixel values in an image).
Training
Labels
The process of feeding the model labeled data
The output or target variable the model
to learn patterns and relationships.
predicts (e.g., "spam" or "not spam," house
prices).
Testing
Evaluating the model's performance on
Model
unseen data to measure its accuracy and
A mathematical representation of the
generalization.
relationship between features and labels that
is used to make predictions.
15
Overfitting and Underfitting
Overfitting
Underfitting
The model learns the training data too well,
The model is too simple and fails to capture
including noise and outliers, which leads to
the underlying patterns in the training data.
poor performance on new, unseen data.
Indicator Indicator: Low accuracy on both training
High accuracy on training data, but low and testing data.
accuracy on testing data.
Solution: Use regularization techniques, Solution: Increase model complexity or use
reduce model complexity, or gather more more features.
data.
16
Data Processing
Data Cleaning
This means removing or correcting errors and inconsistencies in the data to improve model performance.
Examples: Handling missing values, removing duplicates, correcting incorrect data entries.
Data Augmentation
This means creating new, synthetic data from the existing dataset to increase diversity and improve generalization.
Examples: Rotating or flipping images in computer vision tasks, adding noise to time-series data.
Splitting Data into Training, Validation, and Testing Sets
It involves dividing the dataset into separate subsets for training the model, tuning hyperparameters, and
evaluating performance.
Typical Split: 70-80% training, 10-15% validation, 10-15% testing.
Normalization and Standardization
Normalization means scaling data to a specific range, often [0,1].
Standardization means transforming data to have zero mean and unit variance.
17
Popular Machine Learning Frameworks
18
Evaluation Metrics
Accuracy Recall
The proportion of correct predictions out of all The proportion of actual positives correctly identified
by the model
predictions.
Use Case: Useful when the cost of false negatives is
high (e.g., medical diagnosis).
Precision
The proportion of positive predictions that are
actually correct. F1 Score
Use Case: Useful when the cost of false positives is The harmonic mean of precision and recall, providing a
balance between the two.
high (e.g., email spam detection)
19
Ethical Issues in Machine Learning
• Bias in Algorithms: Algorithms can inherit biases present in the training data,
leading to unfair outcomes.
• Data Privacy Concerns: The collection and use of personal data raise concerns
about user privacy and data security.
• Fairness and Accountability: Ensuring that machine learning models make fair
decisions and are accountable for their outcomes.
• Ethical AI Usage: Promoting responsible development and deployment of AI
systems that benefit society.
20
Future of Machine Learning
Advancements in Autonomous Systems
Machine learning is expected to drive the next generation of self-driving cars, drones, and robots.
Quantum Machine Learning
The intersection of quantum computing and machine learning promises faster and more efficient
algorithms.
Edge AI
Running machine learning models locally on devices (e.g., smartphones, IoT devices) for real-time
processing and reduced latency.
Enhanced Interpretability and Explainability:
Efforts are being made to make machine learning models more interpretable and explainable to
increase trust and transparency
21
Challenges in Machine Learning
• Lack of Quality Data: Machine learning models require large amounts of high-
quality data, which is often difficult to obtain.
• Computational Cost: Training complex models, especially deep learning models,
can be computationally expensive.
• Interpretability of Models: Many machine learning models, especially deep
learning models, are seen as "black boxes" with little transparency.
• Ethical Concerns: The ethical implications of AI and machine learning need to be
considered to avoid harm and bias.
22
Key Takeaways
• ML is transforming education by making it more personalized, efficient, and
engaging.
• Challenges like bias and privacy concerns must be addressed for ethical
implementation.
• Collaboration among stakeholders is essential for success.
23
RAIL MEDIA HANDLES
24
THANK YOU
FOR YOUR
TIME
