Midterm 1 Flashcards
The model learns the relationship between inputs and outputs by minimizing _____ between predicted and actual values
the difference
Supervised learning in regression tasks involves fitting the data to a _____ line using _____ data to predict an output y=h(x) from a given input x.
Straight, labeled
The gradient descent algorithm is an optimization method used to minimize a cost function by iteratively updating model parameters in the direction of the _____, which is the _____ of the function
Steepest descent, negative gradient
The update step size is controlled by the _____ , and the algorithm continues until convergence, which is typically defined by a sufficiently small change in the cost function
Learning rate
(T/F) Logistic regression can only be used for binary classification problems
False
(T/F) The output of logistic regression is a probability value between 0 and 1
True
(T/F) Logistic regression does not assume any relationship between the input features and the output
False
(T/F) Logistic regression uses the sigmoid function to model the probability of a class
True
(T/F) SVMs aim to find a decision boundary that maximizes the margin between classes
True
(T/F) The SVM cost function can be approximated by piecewise linear functions, though this increases computational complexity
False
(T/F) The data points closest to the decision boundary are called support vectors
True
(T/F) SVMs achieve better general by maximizing the margin of separation
True
(T/F) Overfitting occurs when a model is too complex and captures noise in the training data
True
(T/F)A model that overfits will have high training accuracy but poor test accuracy
True
(T/F) Overfitting typically happens when the model has too few parameters relative to the training data
False
(T/F) Regularization techniques like L1 or L2 can help prevent overfitting by simplifying the model
True
(T/F) A model that over fits will perform well on both training and test data
False
(T/F) Underfitting typically happens when the model has too many parameters relative to the training data
False
(T/F) Underfitting occurs when a model is too simple to capture the underlying patterns in the data
True
(T/F) Underfitting occurs when a model learns only the noise in the training data
False
(T/F) A model that underfits will have both low training accuracy and low test accuracy
True
(T/F) Increasing model complexity can help address underfitting
True
(T/F) Regularization techniques add a penalty term to the model cost function to reduce the risk of overfitting
True
(T/F) The regularization parameter controls the strength of regularization applied to the model
True
(T/F) Small regularization parameters may allow the model to overfit the training data
True
(T/F) Too large regularization parameters may result in undercutting by making the model too simple
True
(T/F) Regularization can help prevent overfitting even when the training data is small, though more data may be needed for optimal test results
True
(T/F) Increasing regularization always improves model performance, regardless of the situation
False
(T/F) Removing irrelevant or redundant features can help prevent overfitting and improve model generalization
True
(T/F) Adding more training data can reduce overfitting and improve generalization, especially for high-variance models
True
(T/F) Adding polynomial features is a good strategy to fix high-variance (overfitting) problems
False
(T/F) Reducing the number of features by removing irrelevant ones is a method of reducing the risk of overfitting in a machine learning model
True
(T/F) Increasing the regularization parameter(λ) is a method of reducing the risk of overfitting in a machine learning model
True
(T/F) Collecting more training data is a method of reducing the risk of overfitting in a machine learning model
True
(T/F) Using a more complex model with higher polynomial features is a method of reducing the risk of overfitting in a machine learning model
True
Formula for accuracy?
(TP + TN) / (TP + FP + TN + FN)
When do we use regression?
When value is continuous, trying to predict value
When do we use classification?
When data is labeled, trying to categorize data
What is the evaluation metric for regression?
MSE, RMSE, R-Squared
What is the evaluation metric for classification?
Accuracy, Precision, Recall, F-1 Score
Formula for precision?
TP/(TP + FP)
Formula for recall?
TP/(TP + FN)
Formula for F1-Score?
2 * [ (Precision * Recall) / (Precision + Recall)
