How to Learn ML?
How to learn Machine Learning?
ML concept varies but for starters, I will add some links which are really helpful to understand the concepts of ML.
On a higher level Machine Learning could be done based on different algorithms as follows
- Supervised
- UnSupervised
- Reinforcement
Key points to understand
- In supervised learning we teach an algorithm based on known correct answers and then classify new data based on the known answers. It has very discreet classification between positive and negative outputs
- Decision trees is basic classification algorithm used for supervised learning https://medium.com/machine-learning-101/chapter-3-decision-trees-theory-e7398adac567. A decision tree classifier divides a given sample space into smaller sub sections of distinct data
- When data is unlabelled supervised learning is not possible
- Unsupervised learning takes place which is first step towards classification of unlabelled data
- K-Mean algorithm, clustering are one of the most commonly used unsupervised learning for exploratory data
- ReInforcement learning is a model where agents takes decision to maximize the output. It has no specific mathematical model to base its decisions
Different Machine Learning Algorithms
Try to understand following algorithms before going forward.
- Support Vector Machines.
- Linear regression.
- Logistic regression.
- Naive Bayes.
- Linear discriminant analysis.
- Decision trees. http://pages.cs.wisc.edu/~jerryzhu/cs540/handouts/dt.pdf
- K-nearest neighbor algorithm.
- Neural Networks (Multilayer perceptron) -https://en.wikipedia.org/wiki/Multilayer_perceptron
A perceptron is an function which takes two or more inputs and gives an output and is dependent on: https://www.youtube.com/watch?v=ntKn5TPHHAk
- Summation of input with given weighted values
- Activation Function logic which adjust weighted values function
- Also, depending on a learning rate the rate of adjustment is defined
Comments
Post a Comment