WBJECA 2025 — Computer PYQ

The primary distinction between these two machine learning paradigms lies in the presence of labels in the training dataset.

• Supervised Learning: The model is trained on a labeled dataset, where the input $x$ is paired with a corresponding output/label $y$. The goal is to learn a mapping function $f$ such that:

  $$y=f(x)$$

  This is commonly used for tasks like classification and regression.

• Unsupervised Learning: The model works with unlabeled data. The goal is to discover the underlying structure or distribution $P(x)$ in the data without explicit guidance. Common techniques include clustering and dimensionality reduction.

Mathematically, if the dataset is represented as $D=\{x_i,y_i{\}}_{i=1}^n$ for supervised learning, the model minimizes an error function between predicted $\hat{y}$ and actual $y$:

$$Error=\sum _{i=1}^{n}L(y_i,f(x_i))$$

In unsupervised learning, we only have $\{x_i{\}}_{i=1}^n$, and the model identifies groups or associations based on features within the data itself.

• (B) is also a technically correct description, but (A) identifies the prerequisite (labeled data) that defines the category of the algorithm.

• (C) is incorrect because both supervised and unsupervised methods can be used for various tasks, and unsupervised learning is not specifically for regression.

• (D) is incorrect because accuracy is highly dependent on the algorithm, data quality, and the nature of the specific problem being solved.