This algorithm computes the score given a GMM and UBM using the Linear Scoring
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.
| Endpoint Name | Data Format | Nature |
|---|---|---|
| comparison_ids | system/array_1d_text/1 | Input |
| keystroke | tutorial/atvs_keystroke/1 | Input |
| probe_client_id | system/text/1 | Input |
| scores | elie_khoury/string_probe_scores/1 | Output |
| Endpoint Name | Data Format | Nature |
|---|---|---|
| template_client_id | system/text/1 | Input |
| template_id | system/text/1 | Input |
| model_template | aythamimm/keystroke_model/6 | Input |
Parameters allow users to change the configuration of an algorithm when scheduling an experiment
| Name | Description | Type | Default | Range/Choices |
|---|---|---|---|---|
| field | string | given_name | given_name, family_name, email, nationality, id_number, all_five |
xxxxxxxxxximport numpydef scaled_distance(in1,in2,in3): dif=[] mean_gal_features=in1 std_gal_features=in2 probe_features=in3 l_features=len(probe_features) std_threshold=0.2*numpy.mean(std_gal_features); for i in range( l_features): if std_gal_features[i] < std_threshold: std_gal_features[i]=std_threshold for i_dif in range(l_features): dif.append((abs(mean_gal_features[i_dif]-probe_features[i_dif]))/std_gal_features[i_dif]) score = 1/numpy.average(dif) return scoreclass Algorithm: def __init__(self): self.templates = None self.field='given_name' def setup(self, parameters): self.field = parameters.get('field', self.field) return True def process(self, inputs, outputs): # retrieve all the templates once if self.templates is None: self.templates = {} group = inputs.groupOf('model_template') while group.hasMoreData(): group.next() template_id = group['template_id'].data.text if self.field == 'given_name': template1=group['model_template'].data.given_name_average template2=group['model_template'].data.given_name_std if self.field == 'family_name': template1=group['model_template'].data.family_name_average template2=group['model_template'].data.family_name_std if self.field == 'email': template1=group['model_template'].data.email_average template2=group['model_template'].data.email_std if self.field == 'nationality': template1=group['model_template'].data.nationality_average template2=group['model_template'].data.nationality_std if self.field == 'id_number': template1=group['model_template'].data.id_number_average template2=group['model_template'].data.id_number_std if self.field == 'all_five': template11=group['model_template'].data.given_name_average template12=group['model_template'].data.given_name_std template21=group['model_template'].data.family_name_average template22=group['model_template'].data.family_name_std template31=group['model_template'].data.email_average template32=group['model_template'].data.email_std template41=group['model_template'].data.nationality_average template42=group['model_template'].data.nationality_std template51=group['model_template'].data.id_number_average template52=group['model_template'].data.id_number_std if self.field != 'all_five': self.templates[template_id] = dict( client_id = group['template_client_id'].data.text, model_average = template1, model_std = template2, ) if self.field == 'all_five': self.templates[template_id] = dict( client_id = group['template_client_id'].data.text, model_average1 = template11, model_std1 = template12, model_average2 = template21, model_std2 = template22, model_average3 = template31, model_std3 = template32, model_average4 = template41, model_std4 = template42, model_average5 = template51, model_std5 = template52, ) # process the probe comparison_ids = inputs['comparison_ids'].data.text data = inputs['keystroke'].data if self.field == 'given_name': f1 = data.holdtime.given_name f2 = data.rplatency.given_name if self.field == 'family_name': f1 = data.holdtime.family_name f2 = data.rplatency.family_name if self.field == 'email': f1 = data.holdtime.email f2 = data.rplatency.email if self.field == 'nationality': f1 = data.holdtime.nationality f2 = data.rplatency.nationality if self.field == 'id_number': f1 = data.holdtime.id_number f2 = data.rplatency.id_number if self.field == 'all_five': f11 = data.holdtime.given_name f12 = data.rplatency.given_name f21 = data.holdtime.family_name f22 = data.rplatency.family_name f31 = data.holdtime.email f32 = data.rplatency.email f41 = data.holdtime.nationality f42 = data.rplatency.nationality f51 = data.holdtime.id_number f52 = data.rplatency.id_number if self.field != 'all_five': feature_vector=[] l_f=len(f1) for i_k in range(l_f): feature_vector.append(float(f1[i_k])) l_f=len(f2) for i_k in range(l_f): feature_vector.append(float(f2[i_k])) probe_features=feature_vector if self.field == 'all_five': feature_vector1=[] l_f=len(f11) for i_k in range(l_f): feature_vector1.append(float(f11[i_k])) l_f=len(f12) for i_k in range(l_f): feature_vector1.append(float(f12[i_k])) feature_vector2=[] l_f=len(f21) for i_k in range(l_f): feature_vector2.append(float(f21[i_k])) l_f=len(f22) for i_k in range(l_f): feature_vector2.append(float(f22[i_k])) feature_vector3=[] l_f=len(f31) for i_k in range(l_f): feature_vector3.append(float(f31[i_k])) l_f=len(f32) for i_k in range(l_f): feature_vector3.append(float(f32[i_k])) feature_vector4=[] l_f=len(f41) for i_k in range(l_f): feature_vector4.append(float(f41[i_k])) l_f=len(f42) for i_k in range(l_f): feature_vector4.append(float(f42[i_k])) feature_vector5=[] l_f=len(f51) for i_k in range(l_f): feature_vector5.append(float(f51[i_k])) l_f=len(f52) for i_k in range(l_f): feature_vector5.append(float(f52[i_k])) probe_features1=feature_vector1 probe_features2=feature_vector2 probe_features3=feature_vector3 probe_features4=feature_vector4 probe_features5=feature_vector5 scores = [] for comparison_id in comparison_ids: template_client_identity = self.templates[comparison_id]['client_id'] if self.field != 'all_five': score=scaled_distance(self.templates[comparison_id]['model_average'],self.templates[comparison_id]['model_std'],probe_features); if self.field == 'all_five': score=[] score.append(scaled_distance(self.templates[comparison_id]['model_average1'],self.templates[comparison_id]['model_std1'],probe_features1)) score.append(scaled_distance(self.templates[comparison_id]['model_average2'],self.templates[comparison_id]['model_std2'],probe_features2)) score.append(scaled_distance(self.templates[comparison_id]['model_average3'],self.templates[comparison_id]['model_std3'],probe_features3)) score.append(scaled_distance(self.templates[comparison_id]['model_average4'],self.templates[comparison_id]['model_std4'],probe_features4)) score.append(scaled_distance(self.templates[comparison_id]['model_average5'],self.templates[comparison_id]['model_std5'],probe_features5)) score=numpy.average(score) scores.append({ 'template_identity': template_client_identity, 'score': score, }) outputs['scores'].write({ 'client_identity': inputs['probe_client_id'].data.text, 'scores': scores }, end_data_index=inputs['probe_client_id'].data_index_end ) return TrueThe 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, a Gaussian Mixture Model (GMM) of the target client and an UBM-GMM, this algorithm computes the scoring using the linear scoring implemented on the `Bob https://www.idiap.ch/software/bob/docs/releases/last/sphinx/html/machine/generated/bob.machine.linear_scoring.html?highlight=linear%20scoring#bob.machine.linear_scoring`_
| Updated | Name | Databases/Protocols | Analyzers | |||
|---|---|---|---|---|---|---|
| aythamimm/aythamimm/btas15_keystroke_experiments/6/BTAS_2015_Kesytroke_Experiment | atvskeystroke/1@A | aythamimm/analyzer_keystroke/70 |
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.