This algorithm is a legacy one. The API has changed since its implementation. New versions and forks will need to be updated.

This algorithm is splittable

Endpoint Groups 1

Algorithms have at least one input and one output. All algorithm endpoints are organized in groups. Groups are used by the platform to indicate which inputs and outputs are synchronized together. The first group is automatically synchronized with the channel defined by the block in which the algorithm is deployed.

Group: main

Endpoint Name	Data Format	Nature
features	system/array_2d_floats/1	Input
classifier	pkorshunov/two-classes-gmm/1	Input
scores	system/float/1	Output

import numpy
import bob.learn.em

def gmm_from_data(data):
    """Loads a bob.learn.em.GMMMachine from a BEAT Data object"""

dim_c, dim_d = data.means.shape
    gmm = bob.learn.em.GMMMachine(dim_c, dim_d)
    gmm.weights = data.weights
    gmm.means = data.means
    gmm.variances = data.variances
    gmm.variance_thresholds = data.variance_thresholds
    return gmm

class Algorithm:

def __init__(self):
        self.gmm_one      = None
        self.gmm_two      = None
        self.features = []

def process(self, inputs, outputs):

# retrieve the classifier (with two GMM models) once
        if self.gmm_one is None:
            two_gmms = inputs['classifier'].data
            self.gmm_one = gmm_from_data(two_gmms.model_one)
            self.gmm_two = gmm_from_data(two_gmms.model_two)

# read the features and project on both GMMs
        projection_one = 0
        projection_two = 0
        features = inputs['features'].data.value
        for feature in features:
            # project the feature on both GMMs, result of projection is a log likelihood
            projection_one += self.gmm_one.log_likelihood(feature)
            projection_two += self.gmm_two.log_likelihood(feature)
        projection_one /= features.shape[0]        
        projection_two /= features.shape[0]    
        
        # compute the score as the difference between two average log-likelihoods
        score = projection_one - projection_two
        # output the projection score
        outputs['scores'].write({
            'value': score
            })
        
        return True

The code for this algorithm in Python
The ruler at 80 columns indicate suggested POSIX line breaks (for readability).
The editor will automatically enlarge to accomodate the entirety of your input
Use keyboard shortcuts for search/replace and faster editing. For example, use Ctrl-F (PC) or Cmd-F (Mac) to search through this box

For a given set of feature vectors and a Gaussian Mixture Models (GMM), this algorithm implements the Maximum-a-posteriori (MAP) estimation (adapting only the means).

Details of MAP estimation can be found in the paper: Reynolds, Douglas A., Thomas F. Quatieri, and Robert B. Dunn. "Speaker verification using adapted Gaussian mixture models." Digital signal processing 10.1 (2000): 19-41. A very good description on how the MAP estimation works can be found in the `Mathematical Monks's <https://www.youtube.com/watch?v=kkhdIriddSI&index=31&list=PLD0F06AA0D2E8FFBA&spfreload=1>`_ YouTube channel.z

This algorithm relies on the `Bob <http://www.idiap.ch/software/bob/>`_ library.

Docutils System Messages

System Message: ERROR/3 (<string>, line 3); backlink

Unknown target name: "mathematical monks's <https://www.youtube.com/watch?v=kkhdiriddsi&index=31&list=pld0f06aa0d2e8ffba&spfreload=1>".

System Message: ERROR/3 (<string>, line 6); backlink

Unknown target name: "bob <http://www.idiap.ch/software/bob/>".

Experiments

Attestation:

Privacy:

Status:

			Updated	Name	Databases/Protocols	Analyzers
				pkorshunov/pkorshunov/isv-asv-pad-fusion-complete/1/asv_isv-pad_gmm-fusion_lr-pa	avspoof/2@physicalaccess_verification,avspoof/2@physicalaccess_verification_spoof,avspoof/2@physicalaccess_antispoofing,avspoof/2@physicalaccess_verify_train_spoof,avspoof/2@physicalaccess_verify_train	pkorshunov/spoof-score-fusion-roc_hist/1

Scientific Python 2.7 (0.1.0) 2

This table shows the number of times this algorithm has been successfully run using the given environment. Note this does not provide sufficient information to evaluate if the algorithm will run when submitted to different conditions.

algorithms pkorshunov gmm-two-models-projection-train 2

Endpoint Groups 1

Group: main

Docutils System Messages

Experiments

algorithms

pkorshunov

gmm-two-models-projection-train

2