Numerical Methods for Machine Learning

What is gradient descent?

• A mathematical equation used to calculate the slope of a line.

• A technique used to maximize the cost function in machine learning models.

• An optimization algorithm used to minimize the cost function in machine learning models.

• A method for finding the global minimum of a function.

How does gradient descent work?

• Gradient descent is a clustering algorithm used to group similar data points together.

• Gradient descent is a feature selection technique used to identify the most important variables in a dataset.

• Gradient descent is an optimization algorithm used to minimize the cost function in machine learning.

• Gradient descent is a classification algorithm used to predict categorical outcomes.

What are the advantages of using gradient descent in machine learning?

• Gradient descent is only applicable to linear relationships.

• Gradient descent cannot handle non-linear relationships.

• Gradient descent is computationally expensive.

• Gradient descent has advantages such as computational efficiency, finding the global minimum, flexibility, incremental updates, and handling non-linear relationships.

What are the limitations of gradient descent?

• Limited applicability to non-linear problems, dependence on feature scaling, and requirement of differentiable cost function.

• Lack of global convergence, difficulty in handling large datasets, and vulnerability to outliers.

• Local minima, sensitivity to initial conditions, and tuning of learning rate.

• Overfitting, slow convergence, and high computational cost.

What is a neural network?

• A neural network is a type of machine learning model that is inspired by the structure and function of the human brain.

• A neural network is a type of computer program that simulates the behavior of neurons.

• A neural network is a type of algorithm that uses statistical techniques to make predictions.

• A neural network is a type of software that analyzes data and identifies patterns.

How does a neural network learn?

• By adjusting the activation functions based on the input data

• By using reinforcement learning algorithms

• By adjusting the weights and biases based on the input data and desired output using backpropagation algorithm.

• By randomly selecting weights and biases

What are the different types of neural networks?

• deep neural networks

• unsupervised neural networks

• feedforward neural networks, recurrent neural networks, convolutional neural networks, and generative adversarial networks

• supervised neural networks

What is backpropagation in neural networks?

• Method to train neural networks by adjusting weights and biases based on error

• Algorithm used to calculate gradients in neural networks.

• Technique to initialize weights and biases in neural networks.

• Method to optimize the activation function in neural networks.

What are the advantages of using neural networks in machine learning?

• Neural networks can learn from large amounts of data, handle complex relationships, and make accurate predictions.

• Neural networks are not suitable for handling unstructured data.

• Neural networks require less computational power compared to other machine learning algorithms.

• Neural networks are not interpretable and lack transparency in decision-making.

What are the limitations of neural networks?

• Large amounts of data for training, overfitting, lack of interpretability, computational complexity

• Limited ability to handle sequential data, sensitivity to initial conditions, lack of transparency in decision-making