MACHINE LEARNING
Topic : Introduction
Machine Learning
How do humans make decisions?
Dengue
Machine Learning
Humans make decisions from past experiences
Machine Learning
What is the semantics of decision making?
Features and Rules
Machine Learning
How do we outsource these to a machine?
Expert Systems
Machine Learning
Challenges faced
Expert Systems
• A huge amount of data to process
• Unknown rules
• Complex rules
Machine Learning
What if a machine could figure out the function if we provide the needed data
and the parameters of that function?
Machine Learning
Machine Learning
Machine Learning - What
• Arthur Samuel first used the term "machine learning" in 1959.
• Machine Learning a subset of Artificial Intelligence is the most popular technique
of predicting the future or classifying information to help people in making
necessary decisions.
• Data is the core backbone of machine learning algorithms. With the help of the
historical data, we are able to create more data by training these machine learning
algorithms.
• Machine Learning combines computer science, mathematics, and statistics.
• Statistics is essential for drawing inferences from the data. Mathematics is useful
for developing machine learning models and finally, computer science is used
for implementing algorithms.
Machine Learning - Why
• Now a days we are witnessing a vast amount of data.
• In order to derive meaningful insights from this data, we need effective computational
algorithms to churn the data and provide us with useful results.
• Machine Learning has become an essential part of modern industry like
medicine, healthcare, manufacturing, banking etc.
• Machine Learning is being utilized everywhere.
• The electronic devices we use, the applications that are part of our everyday life are powered
by powerful machine learning algorithms.
• Machine Learning example – Google is able to provide us with appropriate search results
based on browsing habits.
• Similarly, Netflix is capable of recommending the films or shows that we would want to
watch based on our watch history.
How Machine Learning Works
Six Jars of ML
How Machine Learning Works
Six Jars of ML
Jar 1: Data
It is the information available in different formats. Information is the key to solve any
problem
How Machine Learning Works
Six Jars of ML
Where do I get the data from?
Data Curation
How Machine Learning Works
Six Jars of ML
What do we do with this data ?
Jar 2: Tasks
How Machine Learning Works
Six Jars of ML
What is the mathematical formulation of a task?
Jar 3 : Model
• A model is a prediction of a relation between input data and output data.
• Consider the generalized function
• Here y is the output and x is the input. and f(x) is a good relationship between them
• Now this relationship can be anything starting from a simple linear equation y = mx+c to
an n-th degree polynomial as shown below.
• These mathematical equations that represent the probable relation between input and
output are called Models.
How Machine Learning Works
Six Jars of ML
How do we know which model is better?
Jar 4: Loss Function
• The loss function helps us to determine the model closest to the true relation between
input and the output.
• The loss function can be defined as follows:
Where,
• L = Loss value (Goal is to find the model which produces the least L)
• y = True output
• f(x) = Output curated by the proposed model
How Machine Learning Works
Six Jars of ML
How do we identify parameters of the Model?
Jar 5: Learning Algorithm
How Machine Learning Works
Six Jars of ML
How would you quantify the performance of your system.
Jar 6: Evaluation
• Among the different models implemented one need to find the best model.
• One of the ways to evaluate the model is by the prediction accuracy of the model.
• Accuracy = Correct number of predictions/total number of predictions.
• Precision and recall make up the accuracy.
• Precision relates to the number of times the model gives the right output. While, recall
relates to the number of times the model gives the wrong output.
• The evaluation is always done on test data instead of training data.
How Machine Learning Works
Machine learning Illustration:
Applications of Machine Learning