Supervised Machine Learning: An Overview
Supervised Machine Learning is an approach that trains algorithms using a dataset comprising examples and corresponding models.
This method is described as supervised since it relies on a guiding figure, typically a data scientist or a machine learning engineer, who selects and provides the dataset, known as the training set, for the learning process.
Understanding the Training Set
The training set is a collection of n examples where each example is represented by a vector xi encompassing j features, and is paired with a label yi, which signifies the correct outcome or category.
This dataset equips the system to autonomously categorize cases or solve problems by analyzing patterns and relationships within the data.
Example: Consider a database filled with thousands of digital car images. The system scrutinizes these images to discern regularities and common patterns.
Through examining these examples, the machine develops a classification hypothesis, using inductive reasoning.
The culmination of this process is the formation of a classification model.
With the model in place, the algorithm can apply it to generalize and classify new, unseen examples.
Example: The system might use this model to recognize cars in various photographs that were not part of the original dataset used to train the model.
Ultimately, supervised machine learning is a process where the machine acquires knowledge from a curated set of classified examples provided by the supervisor. The efficacy of this learning is evident in the model's low margin of error when classifying new data.
Supervised ML Classifications
Supervised machine learning systems are generally categorized into two distinct types:
- Classification
- Regression
- Linear Regression
- Lasso and Ridge Regression
- Non-Linear Regression
