What is Machine Learning?
Machine Learning is simply the strategy of making a machine learn from data. When going deep, we can say that Machine Learning is the subset of Artificial Intelligence that enables computers the ability to learn and improve on their own experience without being explicitly programmed. Mainly there are three types of methods in which machine learning algorithms learn. They are...
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Supervised Learning
In a supervised learning process, the training data you feed to the algorithm includes the desired solutions called the labels. This means, there are already some data that consist of the desired answers or output in the dataset itself.
Let's understand the concept of supervised learning with an example, take the case of a five-year-old student, he/she is not likely to understand the subjects without the help of his/her teacher. So there he/she needs a supervisor for learning the subjects. In the same way, our supervised learning algorithm is also like a five-year-old child which cannot learn without the help of a supervisor. So to make a model more efficient, we need to train them continuously with the labeled training data to yield a good result.
After the algorithm learns the rules and patterns of the data, it creates a model which is an algorithmic equation for producing output data with the rules and patterns derived from training data. Here we are giving all labels to the algorithm to predict the outcome. Once the algorithm is well trained with the data it can be launched in the real world.
important supervised learning algorithms:
- Linear Regression
- Logistic Regression
- Support Vector Machines(SVM)
- k-Nearest Neighbors
- Decision Tree & Random Forests
- Neural Networks
Unsupervised Learning
In unsupervised learning, the data patterns are not classified. Instead, the algorithm tries to uncover the hidden patterns in a dataset and create labels. This means the unsupervised learning algorithms are able to classify the overall data into groups of data that are quite similar in their features. Suppose you want to identify which type of customers are mostly attracted to your products, you can classify them into different groups using unsupervised learning algorithms based on their purchasing behavior and can identify which type of customers are most attracted to your product. In industry, unsupervised learning is particularly powerful in fraud detection where the most dangerous attacks are often those yet to be classified. Moreover, it is used in spam filtering, fraudulent transactions, fraudulent online activities, etc.
Important Unsupervised learning algorithms:
- Clustering
-K-Means
-DBSCAN
-Hierarchical Cluster Analysis(HCA)
- Anomaly detection and novelty detection
- One-class-SVM
- Isolation Forest
- Association rule learning
- Apriori
- Eclat
- Visualization and dimensionality reduction
- Principal Component Analysis(PCA)
- Kernal PCA
- Locally-Linear Embedding(LLE)
- t-distributed Stochastic Neighbour Embedding(t-SNE)
Reinforcement Learning
One of the most advanced learning approaches in Machine Learning is Reinforcement learning. Unlike supervised and unsupervised learning, reinforcement learning is a sophisticated learning technique that continuously improves its model by leveraging feedback from previous iterations. The learning system is called an agent which gets rewards for good performance and penalties for bad performance. At each time the machine can understand which type of strategy is the best fit to get the rewards and to keep away from penalties. The method of reinforcement learning is used to train robots how to walk, jump, run, etc.
Conclusion:
Supervised and Unsupervised learning is mostly used in industries to build smart systems. Moreover, these algorithms can also be used to create insights on data that are quite useful for data scientists when working on large datasets. Reinforcement learning on the other hand is used mostly in the field of robotics.
