Fundamentals of Machine Learning
Learn the fundamental concepts of Machine learning.
What is machine learning?
Arthur Samuel defines Machine learning as:
“Machine Learning is a field of study that gives computers the ability to learn without being explicitly programmed.” ~ Arthur Samuel
Traditional programming vs. machine learning
In traditional programming, the algorithm is known and a function is written to produce the output. In Machine learning, you know the input and the output but the algorithm is the unknown that produces the output.
| Traditional Programming | Machine Learning |
|---|---|
| Input data | Take input and output |
| Apply logic to it to produce a result | Figure out the algorithm |
Temperature conversion problem
Suppose you want to write a program that converts celsius to fahrenheit. The input is in celsius and the output is in fahrenheit.
💡 Question: What is the output for the input value of 40?
In traditional programming, a function can be written that takes input in celsius and returns the output in fahrenheit using the formula . Note that the program was explicitly told the instructions to execute.
With Machine learning approach, the input and the output are known and the model learns the relationship between the input and the output.
The temperature conversion problem is a regression problem because we are predicting a continuous output.
Now, let’s discuss the classification problem.
Fraud detection problem
To classify whether the given transaction data is fraudulent or authorized, a model can be trained to learn the relationship between input transaction data and the output category (fraudulent/authorized). The trained model can then be used to identify the output category, given unseen transaction data.
📝 Question: Is it possible to train a Machine to predict the outcome perfectly?
The answer is that it is not possible. The ultimate goal of Machine learning is to make the “best possible” guess that will be useful. It is not utilized to make a perfect guess.
Takeaway
A Machine Learning algorithm takes the input and output and learns the algorithm that finds the relationship between the input and the output.
The input that is given to the model is the features. The output is the label.
The model will continue training until it finds the best relationship between the input and the output.