HMM Classifier

Multiple HMMs can be combined to form a multi-class classifier.

The HMMClassifier can be used to classify:

  • Univariate/multivariate numerical observation sequences, by using GaussianMixtureHMM models.

  • Univariate categorical observation sequences, by using CategoricalHMM models.

To classify a new observation sequence \(O'\), the HMMClassifier works by:

  1. Creating HMMs \(\lambda_1, \lambda_2, \ldots, \lambda_C\) and training each model using the Baum—Welch algorithm on the subset of training observation sequences with the same class label as the model.
  2. Calculating the likelihoods \(\mathbb{P}(O'\ |\ \lambda_1), \mathbb{P}(O'\ |\ \lambda_2), \ldots, \mathbb{P}(O'\ |\ \lambda_C)\) of each model generating \(O'\) using the Forward algorithm.
  3. Scaling the likelihoods by priors \(p(\lambda_1), p(\lambda_2), \ldots, p(\lambda_C)\), producing un-normalized posteriors
    \(\mathbb{P}(O'\ |\ \lambda_c)p(\lambda_c)\) for each class.
  4. Choosing the class represented by the HMM with the highest posterior probability for \(O'\).
    \[\begin{split}c' = \operatorname*{\arg\max}_{c\in\{1,\ldots,C\}}\ p(\lambda_c\ |\ O') = \operatorname*{\arg\\max}_{c\in\{1,\ldots,C\}}\ \mathbb{P}(O'\ |\ \lambda_c)p(\lambda_c)\end{split}\]

These steps are summarized in the diagram below.

HMM Classifier

API reference

Class

HMMClassifier

A classifier consisting of HMMs, each trained independently to recognize sequences of a single class.

Methods

__init__(*[, prior, classes, n_jobs])

Initialize a HMMClassifier.

add_model(model, /, *, label)

Add a single HMM to the classifier.

add_models(models, /)

Add HMMs to the classifier.

fit([X, y, lengths])

Fit the HMMs to the sequence(s) in X.

fit_predict(X, y, *[, lengths])

Fit the model to the sequence(s) in X and predicts outputs for X.

load(path, /)

Load and deserialize a fitted HMM classifier.

predict(X, *[, lengths])

Predict classes for the sequence(s) in X.

predict_log_proba(X, *[, lengths])

Predict log un-normalized posterior probabilities for the sequences in X.

predict_proba(X, *[, lengths])

Predict class probabilities for the sequence(s) in X.

predict_scores(X, *[, lengths])

Predict class scores for the sequence(s) in X.

save(path, /)

Serialize and save a fitted HMM classifier.

score(X, y, *[, lengths, normalize, ...])

Calculate the predictive accuracy for the sequence(s) in X.

class sequentia.models.hmm.classifier.HMMClassifier

A classifier consisting of HMMs, each trained independently to recognize sequences of a single class.

The predicted class for a given observation sequence is the class represented by the HMM which produces the maximum posterior probability for the observation sequence.

Examples

Using a HMMClassifier (with GaussianMixtureHMM models) to classify spoken digits.

import numpy as np
from sequentia.datasets import load_digits
from sequentia.models.hmm import GaussianMixtureHMM, HMMClassifier

# Seed for reproducible pseudo-randomness
random_state = np.random.RandomState(1)

# Fetch MFCCs of spoken digits
data = load_digits()
train_data, test_data = data.split(test_size=0.2, random_state=random_state)

# Create a HMMClassifier using a class frequency prior
clf = HMMClassifier(prior='frequency')

# Add an untrained HMM for each class
for label in data.classes:
    model = GaussianMixtureHMM(random_state=random_state)
    clf.add_model(model, label=label)

# Fit the HMMs by providing training observation sequences for all classes
clf.fit(train_data.X, train_data.y, lengths=train_data.lengths)

# Predict classes for the test observation sequences
y_pred = clf.predict(test_data.X, lengths=test_data.lengths)

As done in the above example, we can provide unfitted HMMs using add_model() or add_models(), then provide training observation sequences for all classes to fit(), which will automatically train each HMM on the appropriate subset of data.

Alternatively, we may provide pre-fitted HMMs and call fit() with no arguments.

 # Create a HMMClassifier using a class frequency prior
 clf = HMMClassifier(prior='frequency')

# Manually fit each HMM on its own subset of data
 for X_train, lengths_train, label for train_data.iter_by_class():
     model = GaussianMixtureHMM(random_state=random_state)
     model.fit(X_train, lengths=lengths_train)
     clf.add_model(model, label=label)

 # Fit the classifier
 clf.fit()
__init__(*, prior=PriorMode.UNIFORM, classes=None, n_jobs=1)

Initialize a HMMClassifier.

Parameters:

  • self (HMMClassifier) –

  • prior (PriorMode | dict[int, float]) –

    Type of prior probability to assign to each HMM.

    • If None, a uniform prior will be used, making each HMM equally likely.

    • If "frequency", the prior probability of each HMM is equal to the fraction of total observation sequences that the HMM was fitted with.

    • If a dict, custom prior probabilities can be assigned to each HMM. The keys should be the label of the class represented by the HMM, and the value should be the prior probability for the HMM.

  • classes (list[int] | None) –

    Set of possible class labels.

    • If not provided, these will be determined from the training data labels.

    • If provided, output from methods such as predict_proba() and predict_scores() will follow the ordering of the class labels provided here.

  • n_jobs (int) –

    Maximum number of concurrently running workers.

    • If 1, no parallelism is used at all (useful for debugging).

    • If -1, all CPUs are used.

    • If < -1, (n_cpus + 1 + n_jobs) are used — e.g. n_jobs=-2 uses all but one.

Return type:

HMMClassifier

add_model(model, /, *, label)

Add a single HMM to the classifier.

Parameters:

  • self (HMMClassifier) –

  • model (BaseHMM) – HMM to add to the classifier.

  • label (int) – Class represented by the HMM.

Returns:

The classifier.

Return type:

HMMClassifier

Notes

All models added to the classifier must be of the same type — either GaussianMixtureHMM or CategoricalHMM.

add_models(models, /)

Add HMMs to the classifier.

Parameters:

  • self (HMMClassifier) –

  • models (dict[int, BaseHMM]) – HMMs to add to the classifier. The key for each HMM should be the label of the class represented by the HMM.

Returns:

The classifier.

Return type:

HMMClassifier

Notes

All models added to the classifier must be of the same type — either GaussianMixtureHMM or CategoricalHMM.

fit(X=None, y=None, *, lengths=None)

Fit the HMMs to the sequence(s) in X.

Parameters:

  • self (HMMClassifier) –

  • X (ndarray[Any, dtype[float64]] | ndarray[Any, dtype[int64]] | None) – Sequence(s).

  • y (ndarray[Any, dtype[int64]] | None) – Classes corresponding to sequence(s) in X.

  • lengths (ndarray[Any, dtype[int64]] | None) –

    Lengths of the sequence(s) provided in X.

    • If None, then X is assumed to be a single sequence.

    • len(X) should be equal to sum(lengths).

Returns:

The fitted classifier

Return type:

HMMClassifier

fit_predict(X, y, *, lengths=None)

Fit the model to the sequence(s) in X and predicts outputs for X.

Parameters:

  • self (ClassifierMixin) –

  • X (ndarray[Any, dtype[float64]] | ndarray[Any, dtype[int64]]) – Sequence(s).

  • y (ndarray[Any, dtype[int64]]) – Outputs corresponding to sequence(s) in X.

  • lengths (ndarray[Any, dtype[int64]] | None) –

    Lengths of the sequence(s) provided in X.

    • If None, then X is assumed to be a single sequence.

    • len(X) should be equal to sum(lengths).

Returns:

Output predictions.

Return type:

numpy.ndarray

classmethod load(path, /)

Load and deserialize a fitted HMM classifier.

Parameters:
Returns:

Fitted HMM classifier.

Return type:

HMMClassifier

See also

save

Serialize and save a fitted HMM classifier.

predict(X, *, lengths=None)

Predict classes for the sequence(s) in X.

Parameters:
Returns:

Class predictions.

Return type:

numpy.ndarray

Notes

This method requires a trained classifier — see fit().

predict_log_proba(X, *, lengths=None)

Predict log un-normalized posterior probabilities for the sequences in X.

Parameters:
Returns:

Log probabilities.

Return type:

numpy.ndarray

Notes

This method requires a trained classifier — see fit().

predict_proba(X, *, lengths=None)

Predict class probabilities for the sequence(s) in X.

Probabilities are calculated as the posterior probability of each HMM generating the sequence.

Parameters:
Returns:

Class membership probabilities.

Return type:

numpy.ndarray

Notes

This method requires a trained classifier — see fit().

predict_scores(X, *, lengths=None)

Predict class scores for the sequence(s) in X.

Scores are calculated as the log posterior probability of each HMM generating the sequence.

Parameters:
Returns:

Class scores.

Return type:

numpy.ndarray

Notes

This method requires a trained classifier — see fit().

save(path, /)

Serialize and save a fitted HMM classifier.

Parameters:
Return type:

None

Notes

This method requires a trained classifier — see fit().

See also

load

Load and deserialize a fitted HMM classifier.

score(X, y, *, lengths=None, normalize=True, sample_weight=None)

Calculate the predictive accuracy for the sequence(s) in X.

Parameters:
Returns:

Predictive accuracy.

Return type:

float

Notes

This method requires a trained classifier — see fit().

classes: list[int] | None

Set of possible class labels.

models: dict[int, BaseHMM]

HMMs constituting the HMMClassifier.

n_jobs: int

Maximum number of concurrently running workers.

prior: PriorMode | dict[int, float]

Type of prior probability to assign to each HMM.