Machine Learning:Overview
Published On 2021/12/28 Tuesday, Singapore
This posts covers overveiw on different types of Machine Learning as well as notations or terminology of terminology.
Types of Machine Learning
Supervised Learning
Supervised learning learns from the labeled data to predict $x$ to $y$.
Input(X) | Output(y) | Application |
---|---|---|
Spam?(0/1) | Spam Filtering | |
Audio | Text Transcripts | Speech Recognition |
English | Spanish | Machine Translation |
ads - user info | click(0/1) | Online Advertising |
Image - rader info | poisition of other cars | Self-driving Car |
Image of Phone | defect(0/1) | Visual inspection |
Regression: When y is a continuous variable - an infinite number of possible outputs.
An example of regression would be House Price Prediction. whe the input/feature($x$) can be the size of hte house, the output/target variable($y$) is the house price.
Classification: when y is a categorical variable - a small or finite number of possible outputs. Example: breast cancer prediction.
Unsupervised Learning
Input(x) | Application | |
---|---|---|
News Article content | Google News | |
DNA | DNA Microarray | |
Customer informtion | Market Segementation |
“Unsupervised Learning is actually just as super as unsupervised Learning”
Unsupervised learning finds the structure or pattern in unlabelled data. Unlabelled data only comes with inputs x but not output labels y algorithm has to find structure in the data.
Clustering
There are different type of clustering methods.
By techniques, clustering methods can be classified as the follows
- Distance-based methods
- Density-based and grid-based methods
- Probabilistic and generative models
- Leveraging dimensionality reduction methods
- High-dimensional clustering
- Scalable techniques for cluster analysis
By data type, clustering methods can be classified as the follows
- numerical data
- categorical data
- text data
- multimedia data
- time series data
- sequences
- stream data
- network data
- uncertain data
Notation and Terminology
- $x$: Input/Feature
- $y$: Output/Feature
- $f(x)$: Function/Model
- $m$: The number of of training examples.