What is Machine learning?

Machine learning is a way of teaching computers to learn on their own, just like how humans learn from experience. Instead of telling the computer exactly what to do step by step, we give it lots of examples and let it figure out patterns by itself.

 

For example, if we show a computer many pictures of cats and dogs and tell it which is which, over time, it will learn to recognize cats and dogs on its own—even when it sees a new picture it has never seen before.

What is the key characteristics of Machine learning?

Self-learning from the input data is the key characteristic of Machine learning. For example, Machine learning algorithms can identify the patterns, learn rules from the unseen data.

 

Machine learning uses data as input to build predictive models, later this model is used to take decisions on unseen data.

 

For example, if a machine learning model is being trained to identify spam messages in emails, the data would be a set of emails that have been tagged with ‘spam’ in the emails. The model would learn to identify the spams in the emails by analyzing the data.

 

Even though machine has the capability to learn by itself, there is a necessity of human programmer intervention to perform below tasks.

1. To feed the data into model
2. Select appropriate algorithm to address given problem and to tweak various settings of this algorithm etc.,

 

Training and test Data

In Machine learning, input data is split into two categories

 

1. Training data: Data that is used to train a Machine learning model is called the training data. Prediction accuracy of the model is depend on both quality and quantity of the data. If we train the model with poor quality data, then model is not able to make accurate predictions.

 

2. Test data: Second split of data is used to test your model.

 

In general, if you have 100 data points, 80 are used to train the model, 20 are used to test the model.

 

Following are the common training practices.

1. Supervised learning: Training data is labelled or tagged with the correct answers and the model is trained to learn the relationship between the input data and the output data.
2. Unsupervised learning: Training data is not labelled and the model has to learn the patterns/rules by itself from the data.
3. Reinforcement learning: Agent learns to behave in an environment by trial and error. In this model, agent is not explicitly programmed with the rules of the environment, but instead learns by observing the consequences of its actions.

Previous                                                    Next                                                    Home