Interview Questions and Answers

An activation function introduces non-linearity to the neural network, allowing it to learn complex patterns. It transforms the input signal into an output signal, enabling the network to model and understand more intricate relationships in the data.

A loss function measures the difference between the predicted output and the actual target. It quantifies the model's performance during training and is minimized during the optimization process. Common loss functions include mean squared error, cross-entropy, and hinge loss.

A shallow neural network has only a small number of hidden layers, typically one or two, while a deep neural network has a larger number of hidden layers. Deep networks are better at capturing complex hierarchical features in data but may require more data and computational resources.

One-hot encoding is a technique to represent categorical variables as binary vectors. Each category is represented by a unique binary value, with all zeros except for the index corresponding to the category, which is set to one. It is commonly used in tasks like natural language processing.

The activation function introduces non-linearity to the neural network, enabling it to learn complex patterns. Common activation functions include sigmoid, hyperbolic tangent (tanh), and rectified linear unit (ReLU). They allow the network to capture and model more intricate relationships in the data.

The softmax function is used in the output layer of a neural network for multi-class classification. It converts raw output scores into probability distributions, ensuring that the sum of the probabilities across all classes is equal to one. It helps in making a probabilistic prediction for each class.

In a regression problem, the goal is to predict a continuous output, such as a numerical value. In a classification problem, the goal is to assign inputs to one of several predefined categories. Regression models predict quantities, while classification models assign labels.