SDS 506: Supervised vs Unsupervised Learning — with Jon Krohn

Classification & Regression

Jon Krohn explores the nuances of classification and regression problems within supervised learning. Classification involves categorizing inputs into specific buckets, such as determining whether movie reviews are positive or negative, a process known as binary classification. Regression, on the other hand, predicts continuous values like house prices based on input data 1.

Our goal with these supervised learning problems, whether they're classification problems or regression problems, is to learn some function that can take in the inputs x in order to approximate the output y.

--- Jon Krohn

Both types of problems require labeled data to train models effectively, highlighting the importance of accurate input-output mapping 1.

   

Input-Output Labeling

In supervised learning, input-output labeling is crucial for model training. Jon Krohn explains that inputs, typically denoted as 'x', are paired with outputs, or 'y', which are the labels the model aims to predict 1. This process often involves human intervention, such as annotating movie reviews as positive or negative, to provide the necessary labeled data.

We have both the input data x as well as these outputs y, that we'd like to estimate.

--- Jon Krohn

While manual labeling can be costly and time-consuming, alternatives like using existing star ratings can streamline the process, offering a more efficient path to obtaining labeled datasets 1.