Coursera - Andrew Ng ML
ToC
Ref
- ML Specialization - https://www.coursera.org/specializations/machine-learning-introduction#courses
Course 1 - *Supervised Machine Learning: Regression and Classification*
Types of algorithm :
- Supervised learning
- Unsupervised learning
- Recommender systems
- Reinforcement learning
Supervised Learning :
- Right answers(data) given
- e.g. Online Ads platform - Input your data and the ad → output the probability of you clicking it.
Support Vector machine
- There is a mathematical trick to allow computer to deal with infinite number of features
Regression Problem
- Trying to predict a continuous (Non-discrete) valued output
Classification Problem
- Discrete valued output
Unsupervised learning :
- Given only the data without label.
- The goal is to find some pattern, structure, or anything interesting about the data.
Clustering algorithm
- Similar to classification BUT it just does clustering data into groups the model found.
- i.e. We don’t know what are the different types, we let the model find it out
- e.g. Google news clustering the news and provide related/similar news recommendation
- e.g. DNA gene clustering, different group of DNA people may have similar behavior
- e.g. Cluster customers into groups for easier & efficient marketing
- e.g. Deeplearning.ai cluster their community users, to know the purpose of users why joining courses, subscribing newsletters, attending events…etc
Anomaly Detection
- Fraud detection in finance industry
Dimensionality Reduction
- Take a large dataset and reduce it into a smaller dataset
- e.g. file compression