When we say TensorFlow is a machine learning or ML library, what does that mean? To know that we need to know what ML is.
Firstly, machine learning is a type of artificial intelligence or AI.
What is artificial intelligence or AI?
Artificial intelligence allows software applications to become more accurate at predicting outcomes.
However, we don’t have to code for that.
That being said, we can guess that machine learning needs data to process. Right?
Otherwise, how will it predict the outcome based on that data?
There are plenty of examples. For example, we can think of fraud detection, or spam filtering.
Overall we can measure the importance of machine learning in many ways.
Machine learning can give an enterprise a view of trends. It can recommend users what to read based on her choices of readings.
Not only that, based on machine learning a company can launch a new product.
How can we learn Machine Learning?
At the very beginning we need to know that there are different types of machine learning.
You might say different approaches.
Why? Because we need different types of algorithms to handle different types of data.
Isn’t it?
As a result, we need to know the approaches first.
Supervised learning
Firstly, we can think of supervised learning.
Let’s see what that means?
In this category, data scientists supply algorithms with labeled training data.
Here finding the correlations is important.
As a result, we specify both. The input and the output of the algorithm.
Unsupervised learning
In this type of machine learning we don’t use any label. Therefore, the algorithm trains on unlabeled data.
In addition, here the algorithm also tries to find the meaningful connections between data sets. We’ll learn them as we progress.
Semi supervised learning
It’s a mixture of the above two approaches.
However, data scientists use mostly labeled training data. Although the model on its own explores the data and develops its own understanding of the data set.
Reinforcement learning
Reinforcement learning differs from the above two approaches.
Why? Because data scientists try to reinforce or teach the machine to complete a task.
In this scenario, there are defined rules that algorithms follow to complete a task.
Finally, although the algorithm decides on its own to which steps to be followed.
Conclusion on machine learning
Have you found any similar pattern among these different approaches?
There is a commonness.
Firstly, we need a data set to feed the machine learning so that it works upon them.
Secondly, in each case, machine learning tries to find the relationship between the data points.
Finally, we will learn each approach as we progress. Therefore stay tuned.
Leave a Reply