Presented by
Aqsa Riaz (22-Arid-4352)
Haleema Sultan (22-Arid-4362)
Malaika Khalid (22-Arid-4370)
Aiyesha Afzaal (22-Arid-4351)
MACHINE LEARNING
MODELS
MACHINE LEARNING MODELS
 Machine Learning models are very powerful resources
that automate multiple tasks and make them more
accurate and efficient. ML handles new data and scales
the growing demand for technology with valuable insight.
It improves the performance over time.
 A model of machine learning is a set of programs that can
be used to find the pattern and make a decision from an
unseen dataset.
Types Of Machine Learning Models:
 Supervised Models
 Un-Supervised Models
 Semi-Supervised Models
 Reinforcement Models
SUPERVISED MODELS
 Supervised learning is the study of algorithms that use
labeled data in which each data instance has a known
category or value to which it belongs. This results in the
model to discover the relationship between the input features
and the target outcome.
 Classification
The classifier algorithms are designed to indicate whether a new
data point belongs to one or another among several predefined
classes.
 Common Classification Algorithms:
1.Logistic Regression
2.Decision Tree
3.K-Nearest Neighbors (KNN)
 Regression
Regression algorithms are about forecasting of a continuous
output variable using the input features as their basis.
 Common Regression Algorithms
1.Linear Regression
2.Polynomial Regression
UNSUPERVISED MODELS
 Unsupervised learning involves a difficult task of working
with data which is not provided with pre-defined categories
or label.
 Clustering
Visualize being given a basket of fruits with no labels on them.
The fruits clustering algorithms are to group them according to
the inbuilt similarities
SEMI-SUPERVISED LEARNING
 Semi-supervised learning is a type of machine learning that
combines labeled and unlabeled data to train models. This
approach is useful when:- Labeled data is scarce or
expensive to obtain- Unlabeled data is abundant and can
provide valuable information
HOW SEMI-SUPERVISED LEARNING WORKS
 Collect labeled and unlabeled data: Gather a small amount
of labeled data and a larger amount of unlabeled data.
 Train a model on labeled data: Use the labeled data to train
a model, such as a classifier or regressor.
 Use unlabeled data to refine the model: Use the unlabeled
data to refine the model, often through techniques such as self-
training or co-training.
 Evaluate and refine the model: Evaluate the model's
performance on a test set and refine it as needed.
EXAMPLES OF SEMI-SUPERVISED LEARNING
 Image classification: Use a small set of labeled images and
a large set of unlabeled images to train a classifier
 Speech recognition: Use a small set of labeled audio
recordings and a large set of unlabeled recordings to train a
speech recognizer.
 Text classification: Use a small set of labeled text
documents and a large set of unlabeled documents to train a
text classifier.
REINFORCEMENT LEARNING
 Reinforcement learning is a type of machine learning that
involves training an agent to take actions in an environment to
maximize a reward signal. This approach is useful when:
- The environment is dynamic or uncertain
- The goal is to learn a complex behavior or policy
HOW REINFORCEMENT LEARNING WORKS
 Define the environment and agent: Define the environment in
which the agent will operate, as well as the agent itself.
 Define the reward signal: Define the reward signal that the agent
will receive for taking actions in the environment.
 Train the agent: Train the agent using a reinforcement learning
algorithm, such as Q-learning or policy gradients.
 Evaluate and refine the agent: Evaluate the agent's performance
and refine it as needed.
EXAMPLES OF REINFORCEMENT LEARNING
 Game playing: Train an agent to play games such as chess, Go,
or video games.
 Robotics: Train an agent to control a robot to perform tasks
such as navigation or manipulation.
 Autonomous vehicles: Train an agent to control an autonomous
vehicle to navigate roads and avoid obstacles.
DEEP LEARNING
 is a type of machine learning that uses artificial neural networks
with many layers to recognize patterns in data. It's particularly
good for working with large datasets, such as images, text, or
sound.
KEY MODELS IN DEEP LEARNING
Artificial Neural Networks (ANNs):
 Inspired by the human brain.
 Made up of connected nodes (neurons) organized in layers.
 Used for tasks like image and speech recognition.
Convolutional Neural Networks (CNNs):
 Best for analyzing images.
 Automatically detect patterns like edges, shapes, or objects in
pictures.
Recurrent Neural Networks (RNNs):
 Designed for sequential data like text or time-series data.
 Have memory to handle previous inputs for better predictions.
Long Short-Term Memory Networks (LSTMs):
 A type of RNN.
 Great at remembering information for a long time, making
them ideal for tasks like language translation.
Generative Adversarial Networks (GANs):
 Consist of two networks: one creates fake data, and the other
detects it.
 Commonly used to create realistic images or videos.
Transformer Models:
 Popular for natural language tasks (like chatbots).
 Understands long-term relationships in text data effectively.
HOW MACHINE LEARNING WORKS
Model Representation:
 Machine learning models are mathematical functions that
map input (data) to output (predictions).
Learning Algorithm:
 Adjusts the model to reduce errors during training by
improving its parameters step-by-step.
Training Data:
 Input data (features) paired with correct answers (labels)
trains the model to recognize patterns.
Objective Function:
 Measures how far the model's predictions are from the
correct answers. The goal is to minimize this error.
Optimization:
 The process of improving the model to achieve the best
performance, often using methods like gradient descent.
Generalization:
 After training, the model is tested on new data to ensure it
performs well on unseen examples.
Final Output:
 Once trained, the model predicts or classifies new data in
real-world applications.
REAL TIME EXAMPLES FOR ML:
 TRAFFIC PREDICTION
 VIRTUAL PERSONAL ASSISTANT
 ONLINE TRANSPORTATION
 SOCIAL MEDIA SERVICES
 EMAIL SPAM FILTERING
 ONLINE FRAUD DETECTION
BEST PROGRAMMING LANGUAGES FOR ML
Some of the best and most commonly used machine learning programs are;
 Python
 java
 С
 C++
 Shell
 R
 JavaScript
 Shell
FUTURE OF MACHINE LEARNING MODELS
 Smarter: Better at understanding and solving problems.
 Faster: Quicker at making decisions and predictions.
 More Accurate: Less mistakes and more correct answers.
 Easier to Use: Anyone can build and use them, not just experts.
 More Secure: Protected from bad people trying to hack them.
NEW AREAS FOR MACHINE LEARNING MODELS
 Healthcare: Helping doctors find new cures and treatments.
 Environment: Assisting in climate change and conservation
efforts.
 Space Exploration: Aiding in the discovery of new planets
and galaxies.
CONCLUSION
 We have a simple overview of some techniques and algorithms
in machine learning. Furthermore, there are more and more
techniques apply machine learning as a solution. In the future,
machine learning will play an important role in our daily life.
THANK YOU !!

MACHINE LEARNING MODELS. pptx

  • 1.
    Presented by Aqsa Riaz(22-Arid-4352) Haleema Sultan (22-Arid-4362) Malaika Khalid (22-Arid-4370) Aiyesha Afzaal (22-Arid-4351) MACHINE LEARNING MODELS
  • 2.
    MACHINE LEARNING MODELS Machine Learning models are very powerful resources that automate multiple tasks and make them more accurate and efficient. ML handles new data and scales the growing demand for technology with valuable insight. It improves the performance over time.  A model of machine learning is a set of programs that can be used to find the pattern and make a decision from an unseen dataset.
  • 3.
    Types Of MachineLearning Models:  Supervised Models  Un-Supervised Models  Semi-Supervised Models  Reinforcement Models
  • 4.
    SUPERVISED MODELS  Supervisedlearning is the study of algorithms that use labeled data in which each data instance has a known category or value to which it belongs. This results in the model to discover the relationship between the input features and the target outcome.
  • 5.
     Classification The classifieralgorithms are designed to indicate whether a new data point belongs to one or another among several predefined classes.  Common Classification Algorithms: 1.Logistic Regression 2.Decision Tree 3.K-Nearest Neighbors (KNN)
  • 6.
     Regression Regression algorithmsare about forecasting of a continuous output variable using the input features as their basis.  Common Regression Algorithms 1.Linear Regression 2.Polynomial Regression
  • 7.
    UNSUPERVISED MODELS  Unsupervisedlearning involves a difficult task of working with data which is not provided with pre-defined categories or label.  Clustering Visualize being given a basket of fruits with no labels on them. The fruits clustering algorithms are to group them according to the inbuilt similarities
  • 8.
    SEMI-SUPERVISED LEARNING  Semi-supervisedlearning is a type of machine learning that combines labeled and unlabeled data to train models. This approach is useful when:- Labeled data is scarce or expensive to obtain- Unlabeled data is abundant and can provide valuable information
  • 9.
    HOW SEMI-SUPERVISED LEARNINGWORKS  Collect labeled and unlabeled data: Gather a small amount of labeled data and a larger amount of unlabeled data.  Train a model on labeled data: Use the labeled data to train a model, such as a classifier or regressor.  Use unlabeled data to refine the model: Use the unlabeled data to refine the model, often through techniques such as self- training or co-training.  Evaluate and refine the model: Evaluate the model's performance on a test set and refine it as needed.
  • 10.
    EXAMPLES OF SEMI-SUPERVISEDLEARNING  Image classification: Use a small set of labeled images and a large set of unlabeled images to train a classifier  Speech recognition: Use a small set of labeled audio recordings and a large set of unlabeled recordings to train a speech recognizer.  Text classification: Use a small set of labeled text documents and a large set of unlabeled documents to train a text classifier.
  • 11.
    REINFORCEMENT LEARNING  Reinforcementlearning is a type of machine learning that involves training an agent to take actions in an environment to maximize a reward signal. This approach is useful when: - The environment is dynamic or uncertain - The goal is to learn a complex behavior or policy
  • 12.
    HOW REINFORCEMENT LEARNINGWORKS  Define the environment and agent: Define the environment in which the agent will operate, as well as the agent itself.  Define the reward signal: Define the reward signal that the agent will receive for taking actions in the environment.  Train the agent: Train the agent using a reinforcement learning algorithm, such as Q-learning or policy gradients.  Evaluate and refine the agent: Evaluate the agent's performance and refine it as needed.
  • 13.
    EXAMPLES OF REINFORCEMENTLEARNING  Game playing: Train an agent to play games such as chess, Go, or video games.  Robotics: Train an agent to control a robot to perform tasks such as navigation or manipulation.  Autonomous vehicles: Train an agent to control an autonomous vehicle to navigate roads and avoid obstacles.
  • 14.
    DEEP LEARNING  isa type of machine learning that uses artificial neural networks with many layers to recognize patterns in data. It's particularly good for working with large datasets, such as images, text, or sound.
  • 15.
    KEY MODELS INDEEP LEARNING Artificial Neural Networks (ANNs):  Inspired by the human brain.  Made up of connected nodes (neurons) organized in layers.  Used for tasks like image and speech recognition. Convolutional Neural Networks (CNNs):  Best for analyzing images.  Automatically detect patterns like edges, shapes, or objects in pictures. Recurrent Neural Networks (RNNs):  Designed for sequential data like text or time-series data.  Have memory to handle previous inputs for better predictions.
  • 16.
    Long Short-Term MemoryNetworks (LSTMs):  A type of RNN.  Great at remembering information for a long time, making them ideal for tasks like language translation. Generative Adversarial Networks (GANs):  Consist of two networks: one creates fake data, and the other detects it.  Commonly used to create realistic images or videos. Transformer Models:  Popular for natural language tasks (like chatbots).  Understands long-term relationships in text data effectively.
  • 17.
    HOW MACHINE LEARNINGWORKS Model Representation:  Machine learning models are mathematical functions that map input (data) to output (predictions). Learning Algorithm:  Adjusts the model to reduce errors during training by improving its parameters step-by-step. Training Data:  Input data (features) paired with correct answers (labels) trains the model to recognize patterns.
  • 18.
    Objective Function:  Measureshow far the model's predictions are from the correct answers. The goal is to minimize this error. Optimization:  The process of improving the model to achieve the best performance, often using methods like gradient descent. Generalization:  After training, the model is tested on new data to ensure it performs well on unseen examples. Final Output:  Once trained, the model predicts or classifies new data in real-world applications.
  • 19.
    REAL TIME EXAMPLESFOR ML:  TRAFFIC PREDICTION  VIRTUAL PERSONAL ASSISTANT  ONLINE TRANSPORTATION  SOCIAL MEDIA SERVICES  EMAIL SPAM FILTERING  ONLINE FRAUD DETECTION
  • 20.
    BEST PROGRAMMING LANGUAGESFOR ML Some of the best and most commonly used machine learning programs are;  Python  java  С  C++  Shell  R  JavaScript  Shell
  • 21.
    FUTURE OF MACHINELEARNING MODELS  Smarter: Better at understanding and solving problems.  Faster: Quicker at making decisions and predictions.  More Accurate: Less mistakes and more correct answers.  Easier to Use: Anyone can build and use them, not just experts.  More Secure: Protected from bad people trying to hack them.
  • 22.
    NEW AREAS FORMACHINE LEARNING MODELS  Healthcare: Helping doctors find new cures and treatments.  Environment: Assisting in climate change and conservation efforts.  Space Exploration: Aiding in the discovery of new planets and galaxies.
  • 23.
    CONCLUSION  We havea simple overview of some techniques and algorithms in machine learning. Furthermore, there are more and more techniques apply machine learning as a solution. In the future, machine learning will play an important role in our daily life.
  • 24.