Based on the example shown in the lecture (Training R²: 0.95, Test R²: 0.61), what does this indicate?, The model is underfitting, The model is overfitting, The model has perfect generalization, The regularization parameter is too high, What does Elastic Net combine?, L1 and L2 penalties, Gradient descent and coordinate descent, Classification and regression, Linear and polynomial features, What are support vectors in an SVM?, All training samples, Points that lie on or within the margin boundaries, Only misclassified points, The centroid of each class, How does SGDClassifier relate to SVMs?, It cannot approximate SVM behavior, It's only for logistic regression, It's slower than standard SVM for all dataset sizes, It can approximate SVM with appropriate loss function (hinge loss), Which regularization technique performs automatic feature selection?, Ridge (L2), Lasso (L1), Neither, Both equally, What does the gradient descent algorithm minimize in linear regression?, The number of iterations, The number of features, The mean squared error loss function, The learning rate, As we increase the regularization parameter λ in Ridge regression, what happens to bias and variance?, Both bias and variance increase, Both bias and variance decrease, Bias increases, variance decreases, Bias decreases, variance increases, You're training with SGD. What's the tradeoff when choosing batch size?, Large batch → noisy gradients, faster; small batch → stable gradients, slower, Small batch → noisy gradients, slower; large batch → accurate gradients, faster, Large batch → better generalization; small batch → worse accuracy, Batch size only affects epochs, not training.

Tryout

de
Imprimare
Încorporează

Clasament

Stilul vizual

Opţiuni

Comutare șablon

)
Restaurare activitate salvată automat: ?