Sampling

Penalizing

Boosting

Ensemble

Linear

Polynomial

Gaussian

Sigmoid

Stochastic Gradient Descent

Adam

Gradient Descent

Adaboost

Feature selection

Feature scaling

Kernel trick

Regularization

To maximize the margin between classes

To minimize the classification error

To regularize the model

To define the optimal hyperplane

When the data is linearly separable

When the data is noisy

When the data is imbalanced

When the data is sparse

To control regularization strength

To control the degree of kernel function

To control the margin width

To control the number of support vectors

To minimize false positives

To minimize false negatives

To maximize margin between classes

To minimize the number of support vectors

It is computationally efficient for large datasets

It performs well with high-dimensional data

It works well with noisy data

It is not affected by imbalanced datasets

It requires more computational resources compared to binary classification

It is less accurate than other multi-class classification algorithms

It requires re-training for each new class added to the problem

It cannot handle non-linearly separable data