Week #3 in Machine Learning

We cover more Supervised learning algorithms

Last week we covered some classification types in ML. This week I looked into more algothirms that are used in supervised learning for classification and regression problems.

source

4. Naïve Bayesian model

• Used for large finite datasets.
• A method of assigning class labels using a direct acyclic graph. The graph consists a parent node and multiple children’s nodes where each bode is assumed to be independent and separate from the parent.
• There are three types of naïve Bayesian models: Multinominal naïve bayes, Bernoulli naïve bayes,Gaussian naïve bayes.

source

• The technique is primarily used in text classification, spam detection and recommendation systems.

5. Random forests

source

• It is an ensemble method.
• Used for both classification and regression purposes.
• “forest” refers to a collection of uncorrelated decision trees. The trees are later merged to minimize variance and make the predictions more accurate.

6. Neural Networks

• They are primarily leverage for deep learning algorithms.
• They process training data by mimicking interconnectivity in the human brain through layers of nodes.

source

• Each node is made up of inputs, weights and bias.
• Neural networks learn mapping function through supervised learning, adjusting based on the loss function though gradient decent.
• They require significant computational resources.

Important Links

1. https://serokell.io/blog/naive-bayes-classifiers
2. https://www.ibm.com/cloud/learn/supervised-learning
3. https://www.upgrad.com/blog/types-of-supervised-learning/
4. https://www.simplilearn.com/10-algorithms-machine-learning-engineers-need-to-know-article
5. https://towardsdatascience.com/supervised-vs-unsupervised-learning-bf2eab13f288
6. https://www.networkworld.com/article/3587131/machine-learning-in-network-management-has-promise-challenges.html
7. https://www.investopedia.com/terms/n/neuralnetwork.asp