BEAT - tutorial/linear_machines

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 2

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: probes

Endpoint Name	Data Format	Nature
probe_projections	system/array_2d_floats/1	Input
comparison_ids	system/array_1d_uint64/1	Input
probe_id	system/uint64/1	Input
probe_client_id	system/uint64/1	Input
scores	tutorial/probe_scores/1	Output

Group: templates

Endpoint Name	Data Format	Nature
template_client_id	system/uint64/1	Input
template_id	system/uint64/1	Input
template_projections	system/array_2d_floats/1	Input

import bob
import numpy
import scipy.spatial

class Algorithm:

def __init__(self):
        self.templates = None

def process(self, inputs, outputs):

# retrieve all the templates once
        if self.templates is None:

self.templates = {}
            group = inputs.groupOf('template_projections')

while group.hasMoreData():
                group.next()

templates_id = group['template_id'].data.value

self.templates[templates_id] = dict(
                    client_id = group['template_client_id'].data.value,
                    projections = group['template_projections'].data.value,
                )

# process the probe
        comparison_ids  = inputs['comparison_ids'].data.value
        projections     = inputs['probe_projections'].data.value

scores = []
        for comparison_id in comparison_ids:
            template_client_identity = self.templates[comparison_id]['client_id']

distances = []

# gets the min distance for all subprobes w.r.t to all subtemplates
            for subprobe in projections:
                distances.append(numpy.min([scipy.spatial.distance.euclidean(k, subprobe)
                                            for k in self.templates[comparison_id]['projections']]))

# gets the min distance and inverts the scores so matching
            # numbers are numerically bigger than non-matching ones
            scores.append({
                'template_identity': self.templates[comparison_id]['client_id'],
                'score': -numpy.min(distances),
            })

if inputs['probe_id'].isDataUnitDone():
            outputs['scores'].write({
                    'client_identity': inputs['probe_client_id'].data.value,
                    'scores': scores
                },
                end_data_index=inputs['probe_id'].data_index_end
            )

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

This algorithm generates comparison scores between test samples and matrix templates. Comparisons are performed using the Euclidean distance. As a score, the negation of the minimum distance between the row test (probe) vectors and the row vectors from the template matrix is returned.

The inputs are:

probe_projections: a two-dimensional arrays of float (64 bits)

representing the (row) test vectors (the number of test vectors corresponds to the number of rows of the matrix)
comparison_ids: a list of id (as an unsigned 64 bit integers) of

of template/models against which to compare the given probe
probe_id: an identifier (as an unsigned 64 bit integers) for the

given probe
probe_client_id: an identifier (as an unsigned 64 bit integers) for

the client from which this probe samples was generated.
template_client_id: an identifier (as an unsigned 64 bit integers) for

the client from which this template was generated.
template_id: an identifier (as an unsigned 64 bit integers) for the

given template
tempalate_projections: a two-dimensional arrays of float (64 bits)

representing the template matrix

The output scores is the corresponding set of score values.

Experiments

Attestation:

Privacy:

Status:

Name	Databases/Protocols	Analyzers
smarcel/tutorial/eigenface_with_preprocessing/1/eigenface-prepro-tutorial-pca15-ter	atnt/1@idiap	tutorial/postperf_iso/1
anjos/tutorial/eigenface_with_preprocessing/1/foorbar	atnt/1@idiap	tutorial/postperf_iso/1
pranayd/pranayd/pranay/1/pranay	atnt/1@idiap	tutorial/postperf_iso/1
pedro/tutorial/eigenface_with_preprocessing/1/eigenface-with-51-components	atnt/1@idiap	tutorial/postperf_iso/1
EderKapisch/EderKapisch/eigenface_preproc/1/eigenface-with-10-components	atnt/1@idiap	tutorial/postperf_iso/1
tutorial/tutorial/eigenface_with_preprocessing/1/eigenface-with-10-components-preproc	atnt/1@idiap	tutorial/postperf_iso/1
AntonioCandia/AntonioCandia/eigenface_preprop/1/eigenface-with-10-components	atnt/1@idiap	tutorial/postperf/1
tutorial/tutorial/eigenface_with_preprocessing/1/eigenface-with-preproc-15	atnt/1@idiap	tutorial/postperf_iso/1
tutorial/tutorial/full_fisherface/1/mobioMale_fisherfaces_50and20comp	mobio/1@male	tutorial/eerhter_postperf_iso/1
tutorial/tutorial/full_eigenface/1/mobioMale_eigenfaces_50comp	mobio/1@male	tutorial/eerhter_postperf_iso/1

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 tutorial linear_machines_scoring 3

Endpoint Groups 2

Group: probes

Group: templates

Experiments

algorithms

tutorial

linear_machines_scoring

3