ML Algorithms Flashcards
ML Algorithms
What does it mean for data to be linearly separable?
There exists a hyperplane that correctly divides the data.
ML Algorithms
What is a training Algorithm?
An algorithm that finds the parameters of a hyperplane.
ML Algorithms
What is a perceptron?
A training algorithm that generates a hyperplane that separates two classes of examples.
ML Algorithms
How does a perceptron operate?
It iterates over the training exmples and updates weight vector w in a way that makes training examples more likely to be correct
ML Algorithms
How is the weight vector updated in a perceptron algorithm?
The weight vector is replaced by the weight vector plus a small multiple of the example.
w’ = w + nx
ML Algorithms
What is the learning rate in a perceptron?
The multiple the example is multiplied by when adjusting the weight vector.
ML Algorithms
What is an Epoch in training algorithms?
One iteration over the whole data set.
ML Algorithms
In simple terms, what is the Nearest Neighbor Classifier?
Returning the label of the training data example that is closest to the example needing classification.
ML Algorithms
How does a perceptron calculate the class of an example?
WT xi + b = class.
ML Algorithms
What are the classes in a Perceptron?
1, -1
ML Algorithms
What does generalization mean in regards to training algorithms?
The ability of your classifier to perform well on unseen test data.
ML Algorithms
What are the pros of perceptrons?
They are simple and easy to implement
ML Algorithms
What are the cons of perceptrons?
They are limited to linear decision boundaries on binary classification problems.
They also contain several hyperparameters(# of epochs and learning rate)
ML Algorithms
