pystruct.models.BinaryClf¶

class pystruct.models.BinaryClf(n_features=None)[source]¶

Formulate standard linear binary SVM in CRF framework.

Inputs x are simply feature arrays, labels y are -1 or 1.

Parameters:	n_features : int or None, default=None Number of features of inputs x. If None, it is inferred from data.

Notes

No bias / intercept is learned. It is recommended to add a constant one feature to the data.

It is also highly recommended to use n_jobs=1 in the learner when using this model. Trying to parallelize the trivial inference will slow the infernce down a lot!

Methods

batch_inference(X, w)
batch_joint_feature(X, Y)
batch_loss(Y, Y_hat)
batch_loss_augmented_inference(X, Y, w[, ...])
continuous_loss(y, y_hat)
inference(x, w[, relaxed])	Inference for x using parameters w.
initialize(X, Y)
joint_feature(x, y)	Compute joint feature vector of x and y.
loss(y, y_hat)
loss_augmented_inference(x, y, w[, relaxed])	Loss-augmented inference for x and y using parameters w.
max_loss(y)

__init__(n_features=None)[source]¶

inference(x, w, relaxed=None)[source]¶

Inference for x using parameters w.

Finds armin_y np.dot(w, joint_feature(x, y)), i.e. best possible prediction.

For a binary SVM, this is just sign(np.dot(w, x) + b))

Parameters:	x : ndarray, shape (n_features,) Input sample features. w : ndarray, shape=(size_joint_feature,) Parameters of the SVM. relaxed : ignored
Returns:	y_pred : int Predicted class label.

joint_feature(x, y)[source]¶

Compute joint feature vector of x and y.

Feature representation joint_feature, such that the energy of the configuration (x, y) and a weight vector w is given by np.dot(w, joint_feature(x, y)).

Parameters:	x : nd-array, shape=(n_features,) Input sample features. y : int Class label, either +1 or -1.
Returns:	p : ndarray, shape (size_joint_feature,) Feature vector associated with state (x, y).

loss_augmented_inference(x, y, w, relaxed=None)[source]¶

Loss-augmented inference for x and y using parameters w.

Minimizes over y_hat: np.dot(joint_feature(x, y_hat), w) + loss(y, y_hat) which is just sign(np.dot(x, w) + b - y)

Parameters:	x : ndarray, shape (n_features,) Unary evidence / input to augment. y : int Ground truth labeling relative to which the loss will be measured. w : ndarray, shape (size_joint_feature,) Weights that will be used for inference.
Returns:	y_hat : int Label with highest sum of loss and score.