Lecture 13 Flashcards
What is clustering in machine learning?
An unsupervised learning method used to group similar data points together.
What is k-means clustering?
An iterative algorithm that assigns data points to k clusters based on their similarity.
How does k-means clustering determine the number of clusters?
The number of clusters (k) is an input parameter chosen by the user.
What is a disadvantage of k-means clustering?
It can get stuck in local minima and is sensitive to outliers.
What is the silhouette coefficient?
A measure between -1 and 1 that evaluates clustering quality by comparing intra-cluster and inter-cluster similarity.
What is an artificial neural network (ANN)?
A computational model inspired by biological neural networks that learns patterns from data.
What is a perceptron?
A type of artificial neuron that makes binary classifications using a weighted sum of inputs and an activation function.
What is the purpose of an activation function in a neural network?
To introduce non-linearity into the model, allowing it to learn complex patterns.
What is the sigmoid activation function?
A function that maps input values to a range between 0 and 1, often used in binary classification.
What is a feedforward neural network?
A type of neural network where data moves in one direction from input to output without cycles.
What is backpropagation?
A learning algorithm that adjusts neural network weights by computing errors and propagating them backward.
What is a loss function in neural networks?
A function that measures the difference between the predicted output and the actual output.
What is gradient descent?
An optimization algorithm used to minimize the loss function by updating model parameters iteratively.
What is stochastic gradient descent (SGD)?
A variant of gradient descent that updates model parameters using a single sample or a small batch at a time.
What is a deep neural network?
A neural network with multiple hidden layers that can learn complex representations.
What is the difference between supervised and unsupervised learning?
Supervised learning uses labeled data, while unsupervised learning finds patterns in unlabeled data.
What is a multilayer perceptron (MLP)?
A type of feedforward neural network with multiple hidden layers.
What is softmax activation used for?
It converts logits into probabilities for multiclass classification.
What is a generative pre-trained transformer (GPT)?
A deep learning model trained on large datasets to generate human-like text.
How is GPT trained?
Through unsupervised pre-training on large text corpora followed by supervised fine-tuning.
What is beam search in GPT models?
A search algorithm that finds the most likely sequence of words when generating text.
Why do neural networks require large amounts of data?
They have many parameters that need to be trained to generalize well to unseen data.
What is overfitting in neural networks?
When a model learns patterns specific to the training data and fails to generalize to new data.
What is the purpose of model selection in clustering?
To choose the best clustering model based on the data and the intended application.
Why is choosing the right number of clusters (k) challenging?
There is no universal method for selecting k, and different values can lead to different interpretations.
