Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 10

4 TYPES OF

MACHINE LEARNING
TYPES OF
machine learning

01 02
Supervised Unsupervised

03 04
Semi-supervised Reinforcements
01 SUPERVISED
• Definition: Supervised learning is a type of machine learning where
the algorithm learns from a labeled dataset, which means it's
provided with input data and corresponding output labels during
training.

Examples:
• Classification:
Spam email detection (categorizing emails as spam or not).
Image classification
(identifying objects in images like cats or dogs).
• Regression:
Predicting house prices based on features like square footage and location.
Forecasting stock prices based on historical data.
01 SUPERVISED
Advantages:
• It can make accurate predictions when ample labeled data is available.
• Well-suited for tasks where the goal is to categorize or predict specific
outcomes.
• Popular and widely-used in various domains like healthcare, finance, and
natural language processing.
Limitations:
• Requires a substantial amount of labeled data for training, which can be costly
and time-consuming to acquire.
• May struggle with complex relationships between inputs and outputs.
• Limited in its ability to handle outliers or noisy data.
02 UNSUPERVISED
• Definition: Unsupervised learning is a machine learning paradigm where the
algorithm learns patterns and structures in data without explicit output labels. It
aims to discover hidden insights or group similar data points together.

• Examples:
Clustering:
⚬ Grouping customers based on their purchasing behavior without any prior
knowledge of customer segments.
Dimensionality Reduction:
⚬ Reducing the number of features (dimensions) in a dataset while preserving
relevant information. Techniques like Principal Component Analysis (PCA)
fall under this category.
02 UNSUPERVISED
Advantages:
• Useful for exploratory data analysis to discover patterns and insights from data.
• Can be applied when labeled data is scarce or expensive to obtain.
• Helps in feature engineering by reducing data dimensionality and noise.
Limitations:
• Results may be less interpretable compared to supervised learning.
• Finding the right number of clusters or structure can be challenging.
• Performance can be highly dependent on the choice of algorithms and
parameters.
03 SEMI-SUPERVISED
Definition:
Semi-supervised learning is a machine learning paradigm that combines elements of both supervised and unsupervised learning. It
leverages a dataset where a portion of the data is labeled (as in supervised learning), while the majority remains unlabeled (as in
unsupervised learning).

Use Cases:
• Text Classification: In natural language processing, a small subset of documents might be labeled, while the majority are
unlabeled.
• Image Segmentation: Identifying objects or regions in images with only a few examples labeled.
• Anomaly Detection: Recognizing rare events or anomalies in data where most instances are normal.
03 SEMI-SUPERVISED
Advantages:
• Efficiency: It can make efficient use of limited labeled data, reducing the need for extensive manual
labeling.
• Improved Performance: By leveraging both labeled and unlabeled data, semi-supervised learning
models can often achieve better performance compared to purely supervised models.
• Cost Reduction: In cases where labeling data is expensive or time-consuming, semi-supervised
learning can significantly reduce costs.
REINFORCEMENT
04
LEARNING
Definition:
• Reinforcement learning is a machine learning paradigm where an agent learns to make
decisions by interacting with an environment. The agent aims to maximize cumulative
rewards through a series of actions, learning from trial and error.

Examples:
• Game Playing: Reinforcement learning has been used to train agents to play games like
chess, Go, or video games. The agent takes actions (moves) in the game, receives
rewards (winning points or scores), and learns to improve its gameplay over time.
• Robotics: In robotics, reinforcement learning can be applied to teach robots how to
perform tasks like picking up objects, walking, or flying drones. The robot explores its
actions in the physical environment, gets feedback through sensors, and optimizes its
behavior to achieve objectives.
REINFORCEMENT
04
LEARNING
Advantages:
• Suitable for scenarios with sequential decision-making, where actions affect future
outcomes.
• Capable of learning optimal strategies in dynamic and uncertain environments.
• Versatile and applicable in various domains, including robotics, gaming, and
autonomous systems.

Limitations:
• Can require a large number of interactions with the environment, making it
computationally expensive.
• Prone to issues like exploration vs. exploitation trade-offs, where agents need to
balance trying new actions with exploiting known good ones.
• Training can be unstable, especially in complex environments, and may require careful
tuning and regularization techniques.

You might also like