/home/travis/MatthewGerber/asymmetric-threat-tracker/ApplicationData
-
-
- /home/travis/MatthewGerber/asymmetric-threat-tracker/Data/Shapefiles
-
/usr/bin/emacs
-
- /home/travis/MatthewGerber/asymmetric-threat-tracker/Data/Crimes
-
-
-
- /home/travis/MatthewGerber/asymmetric-threat-tracker/Data/Events
-
-
-
- /home/travis/MatthewGerber/asymmetric-threat-tracker/Importers
-
-
diff --git a/Installer/Installer.isl b/Installer/Installer.isl
index e46b1a3..587dd47 100644
--- a/Installer/Installer.isl
+++ b/Installer/Installer.isl
@@ -1923,28 +1923,28 @@
ISBuildSourcePath
ISAttributes
ISComponentSubFolder_
- att_config.xml | ISX_DEFAULTCOMPONENT | ATT_CO~1.XML|att_config.xml | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\ATT\att_config.xml | 1 | |
- building_violations_jan_mar_ | ISX_DEFAULTCOMPONENT3 | BUILDI~1.ZIP|building_violations_jan-mar_2013.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Events\building_violations_jan-mar_2013.zip | 1 | |
- chicago.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago.attimp | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago.attimp | 1 | |
- chicago_bicycle_racks.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Bicycle_Racks.attimp | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Bicycle_Racks.attimp | 1 | |
- chicago_bike_racks.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_bike_racks.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_bike_racks.zip | 1 | |
- chicago_city_boundary.zip | ISX_DEFAULTCOMPONENT6 | CHICAG~1.ZIP|chicago_city_boundary.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\AO\chicago_city_boundary.zip | 1 | |
- chicago_crime_dec_2012_to_ma | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Crime_Dec_2012_to_Mar_2013.attimp | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Crime_Dec_2012_to_Mar_2013.attimp | 1 | |
- chicago_crimes_december_2012 | ISX_DEFAULTCOMPONENT2 | CHICAG~1.ZIP|chicago_crimes_december_2012-march_2013.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Crimes\chicago_crimes_december_2012-march_2013.zip | 1 | |
- chicago_crimes_simple_format | ISX_DEFAULTCOMPONENT2 | CHICAG~1.XML|chicago_crimes_simple_format_sample.xml | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Crimes\chicago_crimes_simple_format_sample.xml | 1 | |
- chicago_major_streets.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Major_Streets.attimp | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Major_Streets.attimp | 1 | |
- chicago_major_streets.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_major_streets.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_major_streets.zip | 1 | |
- chicago_police_stations.atti | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Police_Stations.attimp | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Police_Stations.attimp | 1 | |
- chicago_police_stations.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_police_stations.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_police_stations.zip | 1 | |
+ att_config.xml | ISX_DEFAULTCOMPONENT | ATT_CO~1.XML|att_config.xml | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\ATT\att_config.xml | 1 | |
+ building_violations_jan_mar_ | ISX_DEFAULTCOMPONENT3 | BUILDI~1.ZIP|building_violations_jan-mar_2013.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Events\building_violations_jan-mar_2013.zip | 1 | |
+ chicago.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago.attimp | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago.attimp | 1 | |
+ chicago_bicycle_racks.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Bicycle_Racks.attimp | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Bicycle_Racks.attimp | 1 | |
+ chicago_bike_racks.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_bike_racks.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_bike_racks.zip | 1 | |
+ chicago_city_boundary.zip | ISX_DEFAULTCOMPONENT6 | CHICAG~1.ZIP|chicago_city_boundary.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\AO\chicago_city_boundary.zip | 1 | |
+ chicago_crime_dec_2012_to_ma | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Crime_Dec_2012_to_Mar_2013.attimp | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Crime_Dec_2012_to_Mar_2013.attimp | 1 | |
+ chicago_crimes_december_2012 | ISX_DEFAULTCOMPONENT2 | CHICAG~1.ZIP|chicago_crimes_december_2012-march_2013.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Crimes\chicago_crimes_december_2012-march_2013.zip | 1 | |
+ chicago_crimes_simple_format | ISX_DEFAULTCOMPONENT2 | CHICAG~1.XML|chicago_crimes_simple_format_sample.xml | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Crimes\chicago_crimes_simple_format_sample.xml | 1 | |
+ chicago_major_streets.attimp | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Major_Streets.attimp | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Major_Streets.attimp | 1 | |
+ chicago_major_streets.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_major_streets.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_major_streets.zip | 1 | |
+ chicago_police_stations.atti | ISX_DEFAULTCOMPONENT12 | CHICAG~1.ATT|Chicago_Police_Stations.attimp | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Importers\Chicago_Police_Stations.attimp | 1 | |
+ chicago_police_stations.zip | ISX_DEFAULTCOMPONENT7 | CHICAG~1.ZIP|chicago_police_stations.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Chicago\Features\chicago_police_stations.zip | 1 | |
gui.primary_output | GUI.Primary_output | GUI.Primary output | 0 | | | | 1 | <GUI>|Built | 3 | |
- gui_config.xml | ISX_DEFAULTCOMPONENT | GUI_CO~1.XML|gui_config.xml | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\GUI\gui_config.xml | 1 | |
- kabol_hydro_aquedctl.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_hydro_aquedctl.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_hydro_aquedctl.zip | 1 | |
- kabol_hydro_daml.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_hydro_daml.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_hydro_daml.zip | 1 | |
- kabol_trans_roadl.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_trans_roadl.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_trans_roadl.zip | 1 | |
- kabul_city_boundary.zip | ISX_DEFAULTCOMPONENT10 | KABUL_~1.ZIP|kabul_city_boundary.zip | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\AO\kabul_city_boundary.zip | 1 | |
- license.txt | ISX_DEFAULTCOMPONENT9 | LICENSE.txt | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\LICENSE.txt | 1 | |
- notice.txt | ISX_DEFAULTCOMPONENT9 | NOTICE.txt | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\NOTICE.txt | 1 | |
- readme.txt | ISX_DEFAULTCOMPONENT | README.txt | 0 | | | | 1 | C:\Users\matt\Documents\GitHub\asymmetric-threat-tracker\GUI\Config\README.txt | 1 | |
+ gui_config.xml | ISX_DEFAULTCOMPONENT | GUI_CO~1.XML|gui_config.xml | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\GUI\gui_config.xml | 1 | |
+ kabol_hydro_aquedctl.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_hydro_aquedctl.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_hydro_aquedctl.zip | 1 | |
+ kabol_hydro_daml.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_hydro_daml.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_hydro_daml.zip | 1 | |
+ kabol_trans_roadl.zip | ISX_DEFAULTCOMPONENT11 | KABOL_~1.ZIP|kabol_trans_roadl.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\Features\kabol_trans_roadl.zip | 1 | |
+ kabul_city_boundary.zip | ISX_DEFAULTCOMPONENT10 | KABUL_~1.ZIP|kabul_city_boundary.zip | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\Data\Shapefiles\Kabul\AO\kabul_city_boundary.zip | 1 | |
+ license.txt | ISX_DEFAULTCOMPONENT9 | LICENSE.txt | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\LICENSE.txt | 1 | |
+ notice.txt | ISX_DEFAULTCOMPONENT9 | NOTICE.txt | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\NOTICE.txt | 1 | |
+ readme.txt | ISX_DEFAULTCOMPONENT | README.txt | 0 | | | | 1 | C:\Users\msg8u\Documents\GitHub\asymmetric-threat-tracker\GUI\Config\README.txt | 1 | |
@@ -2690,7 +2690,7 @@
Order
ISSetupLocation
ISReleaseFlags
- _1C35338C_ABD5_46B8_95D3_2ED89AC8618E_ | Microsoft .NET Framework 4.5 Full.prq | | 1 | |
+ _1C35338C_ABD5_46B8_95D3_2ED89AC8618E_ | Microsoft .NET Framework 4.5 Full.prq | | 2 | |
@@ -4457,7 +4457,7 @@ UwBpAG4AZwBsAGUASQBtAGEAZwBlAAEARQB4AHAAcgBlAHMAcwA=
PROGMSG_IIS_ROLLBACKWEBSERVICEEXTENSIONS | ##IDS_PROGMSG_IIS_ROLLBACKWEBSERVICEEXTENSIONS## | |
ProductCode | {55543AF0-BB2E-4BC0-A41E-3370CA7C8F2C} | |
ProductName | Asymmetric Threat Tracker | |
- ProductVersion | 2.0.0 | |
+ ProductVersion | 2.1.0 | |
ProgressType0 | install | |
ProgressType1 | Installing | |
ProgressType2 | installed | |
diff --git a/Installer/Installer.isproj b/Installer/Installer.isproj
index 0f4b724..07f59e3 100644
--- a/Installer/Installer.isproj
+++ b/Installer/Installer.isproj
@@ -1,5 +1,5 @@
-
+
Express
diff --git a/LICENSE.txt b/LICENSE.txt
index 75b5248..7b9703d 100644
--- a/LICENSE.txt
+++ b/LICENSE.txt
@@ -187,7 +187,7 @@
same "printed page" as the copyright notice for easier
identification within third-party archives.
- Copyright [yyyy] [name of copyright owner]
+ Copyright 2014 The Rector & Visitors of the University of Virginia
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
diff --git a/Libraries/FeatureSelector.exe b/Libraries/FeatureSelector.exe
index 6682018..f0bb27b 100644
Binary files a/Libraries/FeatureSelector.exe and b/Libraries/FeatureSelector.exe differ
diff --git a/Libraries/FeatureSelector.xml b/Libraries/FeatureSelector.xml
index 24605de..39e76f6 100644
--- a/Libraries/FeatureSelector.xml
+++ b/Libraries/FeatureSelector.xml
@@ -4,109 +4,162 @@
FeatureSelector
-
+
- Abstract wrapper for subset selection
+ Specifies the cross-validation setup
-
+
- Initializes this wrapper with various things
+ Constructor
+
+
+
+
+ Initializes options
- ID of subset evaluation
- Features to evaluate
Options and their values
-
+
- Initializes subclasses with options
+ Initializes options
- Options to initialize with
+ Options and their values
+ Movement of training instances if randomizing or null if no randomization is performed
-
+
- Sets the gain for a feature
+ Modifies the validation scorer - called immediately before subsets are evaluated
- Feature to set gain for
- Gain of feature
+ Start number of validation instances
+ Length of validation instances
+ Scorer to use
-
+
- Gets the gain of a feature
+ Cleans up any resources created/held by this validator
- Feature to get gain for
- Feature gain
-
+
- Gets whether or not this wrapper contains a given feature (group)
+ Gets the number of folds to use (default: 1, i.e., no cross-validation)
- Feature (group) to check for
- True if feature (group) is contained, false otherwise
-
+
- Gets predicted labels and their scores
+ Gets the block sizes for the instances
- Training vectors
- Validation vectors
- Feature number groups
- Predicted labels
- Prediction scores
-
+
- Cleans up after subset training and validation is complete
+ CrossFoldValidator options
-
+
- Gets the ID of this subset
+ Seed value to use for random number generator that randomizes instance blocks. ([int.MinValue,int.MaxValue], default: picked by Random() constructor)
- ID
-
+
- Gets enumerator over feature vectors, where each dimension in the vector is the feature's value for an instance
+ Path to file that defines the instance block sizes. (optional)
+
+
+
+
+ Number of folds to validate. ([1,int.MaxValue], default: 10)
+
+
+
+
+ Whether or not to randomize training instances before performing cross-validation. This is essential in situations where instance features or class labels are somehow dependent on the instances' positions in the training instance file. The word "blocks" is used due to the capability of treating groups of instances as a single unit; however, unless InstanceBlocksPath is defined, each training instance is treated as its own block. (true/false, default: false)
+
+
+
+
+ Filters features using cosine similarity
+
+
+
+
+ Performs feature filtering
+
+
+
+
+ Initializes this filter
+
+
+
+
+
+ Initializes this filter
+
+
+
+
+
+ Filters a set of features
Feature vectors
Total number of instances
Feature number group mapping
- Mapping from feature ids to instance numbers to feature values
+ Features to remove
-
+
- Gets option configuration for evaluating interaction features
+ Filters features
-
-
-
+
+
- New options
+
+
-
+
- Gets the ID
+ Filters a set of features
+ Feature vectors
+ Total number of instances
+ Feature number group mapping
+ Features removed by previous filters in the chain
+ Features to remove
-
+
- Gets ordered list of features
+ Sets the next filter in the filter chain
-
+
- Gets the newest feature added to the current wrapper
+ Gets the wrapper type
-
+
- Gets or sets the gain of the newest feature added to the current wrapper
+ Initializes options
+
-
+
- Gets or sets the score
+ Filters features
+
+ Feature vectors
+ Total number of instances
+ Feature number group mapping
+ Features already removed
+ Features numbers to remove
+
+
+
+ Options
+
+
+
+
+ Threshold to use when filtering features. ([0,1], default: 0.95)
@@ -223,194 +276,240 @@
Option to get value for
Value for option
-
+
- Filters features using cosine similarity
+ Scores predictions for accuracy
-
+
- Performs feature filtering
+ Abstract base class for all scorers
-
+
- Initializes this filter
+ Constructor
-
-
+
- Initializes this filter
+ Initializes this scorer
-
+ Training instance comments
+ Validation instance comments
+ Option values
-
+
- Filters a set of features
+ Initializes options
- Feature vectors
- Total number of instances
- Feature number group mapping
- Features to remove
+ Options and their values
-
+
- Filters features
+ Gets score for predicted labels
-
-
-
-
-
+ Start of instance range
+ End of instance range
+ Predicted labels
+ Prediction scores
+ True labels
+ Other information about the scoring operation that should be logged
+ Score of predicted labels given true labels
-
+
- Filters a set of features
+ Gets the size of the block that starts at a given instance
- Feature vectors
- Total number of instances
- Feature number group mapping
- Features removed by previous filters in the chain
- Features to remove
+ Number of instance on which the block starts
+ Size of block
-
+
- Sets the next filter in the filter chain
+ Gets or sets the training instance comments
-
+
- Gets the wrapper type
+ Gets or sets the block sizes for the validation instances that are being scored. This will only be non-null
+ when doing cross-validation, as that is the only place that block sizes are applicable.
-
+
+
+ Gets or sets the validation instance comments
+
+
+
+
+ Initializes the options for this scorer
+
+
+
+
+
+ Gets accuracy score for predictions
+
+ Start of instance range
+ End of instance range
+ Predicted labels
+ Prediction scores (not used)
+ True labels
+ Extra information to log (output)
+ Accuracy score
+
+
+
+ Scores predictions with area under surveillance plot
+
+
+
Initializes options
-
+
- Filters features
+ Gets surveillance score
- Feature vectors
- Total number of instances
- Feature number group mapping
- Features already removed
- Features numbers to remove
+ Start of instance range
+ End of instance range
+
+
+
+
+
-
+
Options
-
+
- Threshold to use when filtering features. ([0,1], default: 0.95)
+ A space-separated list of classes to ignore when computing the surveillance plot. (optional)
-
+
- Scores predictions with area under surveillance plot
+ Performs feature selection using SvmLight
-
+
+
+ Wrapper for classifiers such as SvmLight and LibLinear, which use separate training and classification
+ executables. This also assumes that all instance files have lines in the following format:
+
+ label feature1:value1 feature2:value2 ... featureN:valueN
+
+ The one-based feature numbers are sorted ascendingly left to right.
+
+
+
+
+ Abstract wrapper for subset selection
+
+
+
+
+ Initializes this wrapper with various things
+
+ ID of subset evaluation
+ Features to evaluate
+ Options and their values
+
+
- Abstract base class for all scorers
+ Initializes subclasses with options
+ Options to initialize with
-
+
- Constructor
+ Sets the gain for a feature
+ Feature to set gain for
+ Gain of feature
-
+
- Initializes this scorer
+ Gets the gain of a feature
- Training instance comments
- Validation instance comments
- Option values
+ Feature to get gain for
+ Feature gain
-
+
- Initializes options
+ Gets whether or not this wrapper contains a given feature (group)
- Options and their values
+ Feature (group) to check for
+ True if feature (group) is contained, false otherwise
-
+
- Gets score for predicted labels
+ Gets predicted labels and their scores
- Start of instance range
- End of instance range
+ Training vectors
+ Validation vectors
+ Feature number groups
Predicted labels
Prediction scores
- True labels
- Other information about the scoring operation that should be logged
- Score of predicted labels given true labels
-
+
- Gets the size of the block that starts at a given instance
+ Cleans up after subset training and validation is complete
- Number of instance on which the block starts
- Size of block
-
+
- Gets or sets the training instance comments
+ Gets the ID of this subset
+ ID
-
+
- Gets or sets the block sizes for the validation instances that are being scored. This will only be non-null
- when doing cross-validation, as that is the only place that block sizes are applicable.
+ Gets enumerator over feature vectors, where each dimension in the vector is the feature's value for an instance
+ Feature vectors
+ Total number of instances
+ Feature number group mapping
+ Mapping from feature ids to instance numbers to feature values
-
+
- Gets or sets the validation instance comments
+ Gets option configuration for evaluating interaction features
+
+
+
+
+ New options
-
+
- Initializes options
+ Gets the ID
-
-
+
- Gets surveillance score
+ Gets ordered list of features
- Start of instance range
- End of instance range
-
-
-
-
-
-
+
- Options
+ Gets the newest feature added to the current wrapper
-
+
- A space-separated list of classes to ignore when computing the surveillance plot. (optional)
+ Gets or sets the gain of the newest feature added to the current wrapper
-
-
- Wrapper for classifiers such as SvmLight and LibLinear, which use separate training and classification
- executables. This also assumes that all instance files have lines in the following format:
-
- label feature1:value1 feature2:value2 ... featureN:valueN
-
- The one-based feature numbers are sorted ascendingly left to right.
-
+
+
+ Gets or sets the score
+
@@ -528,6 +627,62 @@
Whether or not to include zero-vectors in the validation data. (true/false, default: true)
+
+
+ Constructor
+
+
+
+
+ Initializes this wrapper with options
+
+ Options and their values
+
+
+
+ Gets predicted labels and their scores
+
+ Training vectors
+ Validation vectors
+ Feature number groups
+ Predicted labels
+ Prediction scores
+
+
+
+ Gets the training process arguments.
+
+
+
+
+ Gets the validation process arguments
+
+
+
+
+ SvmLightWrapper options
+
+
+
+
+ Tradeoff between training error and margin. ([float.MinValue, float.MaxValue], default: [avg. x*x]^-1)
+
+
+
+
+ Cost factor, by which training errors on positive examples outweigh errors on negative examples. ([float.MinValue, float.MaxValue], default: 1)
+
+
+
+
+ Fraction of unlabeled exapmles to be classified into the positive class. ([0, 1], default: ratio in labeled data)
+
+
+
+
+ Size of cache for kernel evaluations in MB. ([5, int.MaxValue], default: 40)
+
+
Performs feature selection using LibLinear
@@ -683,14 +838,107 @@
-w13 option in LibLinear training. ([float.MinValue, float.MaxValue], default: 1)
-
+
+
+ -w14 option in LibLinear training. ([float.MinValue, float.MaxValue], default: 1)
+
+
+
+
+ -w15 option in LibLinear training. ([float.MinValue, float.MaxValue], default: 1)
+
+
+
+
+ Treats each validation instance independently. Scores f-measure. Tunes threshold.
+
+
+
+
+ Gets the best confusion matrix and score
+
+ Matrixes to test
+ Labels to compute f-measure with
+ Whether or not to macro-average f-measures
+ Best confusion matrix (output)
+ Best score (output)
+
+
+
+ Constructor
+
+
+
+
+ Initializes options
+
+ Options and their values
+
+
+
+ Scores predictions
+
+ Start of instance range
+ End of instance range
+ Predicted labels
+ Prediction scores
+ True labels
+ Miscellaneous information to log
+ Score of predicted labels given true labels
+
+
+
+ Gets threshold matrixes
+
+ Thresholded matrixes
+
+
+
+ Adds a true/predicted label pair to thresholded prediction matrixes, using the no-judgment class if needed.
+
+ Predicted label
+ Prediction score
+ True label
+ Matrixes to add labels to
+
+
+
+ Gets labels used for f-measure
+
+
+
+
+ Options
+
+
+
+
+ Whether or not to do block scoring during cross-validation. In block scoring, a block (or group) of instances is classified and the highest scoring instance is selected for scoring (other instances are ignored). This only applies to cross-validation setups in which InstanceBlocksPath is supplied to the cross-validator. (true/false, default: false)
+
+
+
+
+ Path to file that lists all possible labels output by the classifier. (mandatory)
+
+
+
+
+ Threshold increment when doing threshold tuning. ((0, float.MaxValue], default: 0.01)
+
+
+
+
+ Maximum threshold when doing threshold tuning. ([float.MinValue, float.MaxValue], default: 1)
+
+
+
- -w14 option in LibLinear training. ([float.MinValue, float.MaxValue], default: 1)
+ Minimum threshold when doing threshold tuning. ([float.MinValue, float.MaxValue], default: 0)
-
+
- -w15 option in LibLinear training. ([float.MinValue, float.MaxValue], default: 1)
+ Whether or not to use macro-averaging for multiple classes. (true/false, default: false)
@@ -1007,77 +1255,6 @@
Type name of the classification wrapper to use. (mandatory)
-
-
- Specifies the cross-validation setup
-
-
-
-
- Constructor
-
-
-
-
- Initializes options
-
- Options and their values
-
-
-
- Initializes options
-
- Options and their values
- Movement of training instances if randomizing or null if no randomization is performed
-
-
-
- Modifies the validation scorer - called immediately before subsets are evaluated
-
- Start number of validation instances
- Length of validation instances
- Scorer to use
-
-
-
- Cleans up any resources created/held by this validator
-
-
-
-
- Gets the number of folds to use (default: 1, i.e., no cross-validation)
-
-
-
-
- Gets the block sizes for the instances
-
-
-
-
- CrossFoldValidator options
-
-
-
-
- Seed value to use for random number generator that randomizes instance blocks. ([int.MinValue,int.MaxValue], default: picked by Random() constructor)
-
-
-
-
- Path to file that defines the instance block sizes. (optional)
-
-
-
-
- Number of folds to validate. ([1,int.MaxValue], default: 10)
-
-
-
-
- Whether or not to randomize training instances before performing cross-validation. This is essential in situations where instance features or class labels are somehow dependent on the instances' positions in the training instance file. The word "blocks" is used due to the capability of treating groups of instances as a single unit; however, unless InstanceBlocksPath is defined, each training instance is treated as its own block. (true/false, default: false)
-
-
Performs feature selection using SvmPerf
@@ -1124,182 +1301,5 @@
Tradeoff between training error and margin (-c in SvmPerf training). ([float.MinValue, float.MaxValue], default: 0.01)
-
-
- Treats each validation instance independently. Scores f-measure. Tunes threshold.
-
-
-
-
- Gets the best confusion matrix and score
-
- Matrixes to test
- Labels to compute f-measure with
- Whether or not to macro-average f-measures
- Best confusion matrix (output)
- Best score (output)
-
-
-
- Constructor
-
-
-
-
- Initializes options
-
- Options and their values
-
-
-
- Scores predictions
-
- Start of instance range
- End of instance range
- Predicted labels
- Prediction scores
- True labels
- Miscellaneous information to log
- Score of predicted labels given true labels
-
-
-
- Gets threshold matrixes
-
- Thresholded matrixes
-
-
-
- Adds a true/predicted label pair to thresholded prediction matrixes, using the no-judgment class if needed.
-
- Predicted label
- Prediction score
- True label
- Matrixes to add labels to
-
-
-
- Gets labels used for f-measure
-
-
-
-
- Options
-
-
-
-
- Whether or not to do block scoring during cross-validation. In block scoring, a block (or group) of instances is classified and the highest scoring instance is selected for scoring (other instances are ignored). This only applies to cross-validation setups in which InstanceBlocksPath is supplied to the cross-validator. (true/false, default: false)
-
-
-
-
- Path to file that lists all possible labels output by the classifier. (mandatory)
-
-
-
-
- Threshold increment when doing threshold tuning. ((0, float.MaxValue], default: 0.01)
-
-
-
-
- Maximum threshold when doing threshold tuning. ([float.MinValue, float.MaxValue], default: 1)
-
-
-
-
- Minimum threshold when doing threshold tuning. ([float.MinValue, float.MaxValue], default: 0)
-
-
-
-
- Whether or not to use macro-averaging for multiple classes. (true/false, default: false)
-
-
-
-
- Scores predictions for accuracy
-
-
-
-
- Initializes the options for this scorer
-
-
-
-
-
- Gets accuracy score for predictions
-
- Start of instance range
- End of instance range
- Predicted labels
- Prediction scores (not used)
- True labels
- Extra information to log (output)
- Accuracy score
-
-
-
- Performs feature selection using SvmLight
-
-
-
-
- Constructor
-
-
-
-
- Initializes this wrapper with options
-
- Options and their values
-
-
-
- Gets predicted labels and their scores
-
- Training vectors
- Validation vectors
- Feature number groups
- Predicted labels
- Prediction scores
-
-
-
- Gets the training process arguments.
-
-
-
-
- Gets the validation process arguments
-
-
-
-
- SvmLightWrapper options
-
-
-
-
- Tradeoff between training error and margin. ([float.MinValue, float.MaxValue], default: [avg. x*x]^-1)
-
-
-
-
- Cost factor, by which training errors on positive examples outweigh errors on negative examples. ([float.MinValue, float.MaxValue], default: 1)
-
-
-
-
- Fraction of unlabeled exapmles to be classified into the positive class. ([0, 1], default: ratio in labeled data)
-
-
-
-
- Size of cache for kernel evaluations in MB. ([5, int.MaxValue], default: 40)
-
-
diff --git a/Libraries/LAIR.Collections.dll b/Libraries/LAIR.Collections.dll
index 76083f4..b9bf3d5 100644
Binary files a/Libraries/LAIR.Collections.dll and b/Libraries/LAIR.Collections.dll differ
diff --git a/Libraries/LAIR.Collections.xml b/Libraries/LAIR.Collections.xml
index 2ad7ba0..2a1cac8 100644
--- a/Libraries/LAIR.Collections.xml
+++ b/Libraries/LAIR.Collections.xml
@@ -4,6 +4,112 @@
LAIR.Collections
+
+
+ Represents an indexed item
+
+
+
+
+ Constructor
+
+ Index of item
+
+
+
+ Gets the index for an item
+
+
+
+
+ Represents an indexable set
+
+ Type of items to store
+ Type of index to use on items
+
+
+
+ Constructor
+
+ Initial capacity
+
+
+
+ Adds an item
+
+ Item to add
+ True if item was new
+
+
+
+ Adds a range of items
+
+ Items to add
+
+
+
+ Gets item by index
+
+ Index of item to fetch
+ Item
+
+
+
+ Tries to get an item for an index
+
+ Index
+ Item
+ True if item was found and false otherwise
+
+
+
+ Removes an item from this set
+
+ Item to remove
+
+
+
+ Gets enumerator over items
+
+
+
+
+
+ Gets enumerator over items
+
+
+
+
+
+ Gets the item for an index
+
+ Index of item to get
+ Item
+
+
+
+ Gets the number of items in this set
+
+
+
+
+ Provides permutations of a sequence of items
+
+ Type of items to permute
+
+
+
+ Constructor
+
+ Items to permute
+ Length of permutation to return (-1 for all permutations)
+
+
+
+ Gets enumerator over permutations of items
+
+ Enumerator over permutations of items
+
Represents a set of unique items
@@ -427,111 +533,5 @@
Current capacity
Next capacity larger than a given capacity
-
-
- Provides permutations of a sequence of items
-
- Type of items to permute
-
-
-
- Constructor
-
- Items to permute
- Length of permutation to return (-1 for all permutations)
-
-
-
- Gets enumerator over permutations of items
-
- Enumerator over permutations of items
-
-
-
- Represents an indexed item
-
-
-
-
- Constructor
-
- Index of item
-
-
-
- Gets the index for an item
-
-
-
-
- Represents an indexable set
-
- Type of items to store
- Type of index to use on items
-
-
-
- Constructor
-
- Initial capacity
-
-
-
- Adds an item
-
- Item to add
- True if item was new
-
-
-
- Adds a range of items
-
- Items to add
-
-
-
- Gets item by index
-
- Index of item to fetch
- Item
-
-
-
- Tries to get an item for an index
-
- Index
- Item
- True if item was found and false otherwise
-
-
-
- Removes an item from this set
-
- Item to remove
-
-
-
- Gets enumerator over items
-
-
-
-
-
- Gets enumerator over items
-
-
-
-
-
- Gets the item for an index
-
- Index of item to get
- Item
-
-
-
- Gets the number of items in this set
-
-
diff --git a/Libraries/LAIR.Extensions.dll b/Libraries/LAIR.Extensions.dll
index c3f6708..e2b7e8d 100644
Binary files a/Libraries/LAIR.Extensions.dll and b/Libraries/LAIR.Extensions.dll differ
diff --git a/Libraries/LAIR.Extensions.xml b/Libraries/LAIR.Extensions.xml
index 81359a3..ed7d054 100644
--- a/Libraries/LAIR.Extensions.xml
+++ b/Libraries/LAIR.Extensions.xml
@@ -4,355 +4,6 @@
LAIR.Extensions
-
-
- Provides extension methods for strings
-
-
-
-
- Constructor
-
-
-
-
- Replaces strings within a string
-
- String to process
- Replacement string pairs, where the key is the string to find and the value is the replacement
- Whether or not to repeat replacement procedure until no changes are made
- String with replacements made
-
-
-
- Removes leading and trailing punctuation from a string
-
- String to trim punctuation from
- Trimmed string
-
-
-
- Removes leading and trailing punctuation from a string
-
- String to trim punctuation from
- Whether or not to trim leading punctuation
- Whether or not to trim trailing punctuation
- Trimmed string
-
-
-
- Removes punctuation characters (any that aren't a-z, A-Z, or 0-9)
-
- String to process
- String without punctuation
-
-
-
- Replaces punctuation characters (any that aren't a-z, A-Z, or 0-9) with something else
-
-
-
-
-
-
-
- Removes all whitespace characters from a string (\s regex character class)
-
- String to process
- String without whitespace
-
-
-
- Removes repeated whitespace from a string
-
- String to process
- String without repeated whitespace
-
-
-
- Throws an exception if any of the given characters are present in the string
-
- String to check
- Character(s) to disallow
-
-
-
- Splits a string on space characters, guaranteeing a specific number of parts. Will throw an exception if the expected number of parts is not found.
-
- String to split
- Number of parts expected
- Parts resulting from split
-
-
-
- Splits a string on space characters, guaranteeing a specific number of parts. Will throw an exception if the expected number of parts is not found.
-
- String to split
- Number of parts expected
- Characters to split on
- Parts resulting from split
-
-
-
- Gets enumeration of parts within a string, delimited by space characters
-
-
-
-
-
-
- Converts a string to its XML-escaped version
-
- Text to convert
- XML-escaped version of text
-
-
-
- Unescapes an string that has been XML-escaped
-
- Text to convert
- Unescaped XML text
-
-
-
- Gets path relative to another path
-
- Base path for absolute path
- Absolute path
- Relative path
-
-
-
- Gets the common initial substring between two strings
-
- First string
- Second string
- Common initial substring
-
-
-
- Changes the first n characters of a string to uppercase
-
- String to change
- Number of characters to change
- Modified string
-
-
-
- Gets the index of the nth non-space character within a string
-
- String to search
- n
- Index of nth non-space character
-
-
-
- Gets the index of the nth occurrence of a character
-
- String to search
- Character to search for
- Value of n
- Index of the nth occurrence of c
-
-
-
- Gets all indexes of a substring within the current string
-
- String to search
- Substring to search for
- Indexes
-
-
-
- Concatenates an enumeration of strings
-
- Strings to concatenate
- Separator for strings
- Concatenated string
-
-
-
- Encrypts a string using AES encryption
-
- String to encrypt
- Encryption key to use
- Initialization to use
- Encrypted bytes
-
-
-
- Decrypts a string using AES encryption
-
- Bytes to decrypt
- Encryption key used to produce the bytes
- Initialization that was used to produce the bytes
- Unencrypted string
-
-
-
- Provides extension methods for TreeBank nodes
-
-
-
-
- Gets the argument index for a node. Node can only have a single index associated with it. An
- exception will be thrown if it has multiple indexes.
-
- Node to get argument index for. Must be a NomBankNode or a PropBankNode
- Whether or not to convert NomBank argument indexes to PropBank indexes when
- possible
- Argument index
-
-
-
- Gets the argument indexes for a node. Node can have multiple indexes, as opposed to GetArgumentIndex. Of course,
- calling this on PropBank nodes will only ever return a single index.
-
- Node to get argument indexes for. Must be a NomBankNode or a PropBankNode.
- Whether or not to convert NomBank argument indexes to PropBank indexes when
- possible
- Argument indexes
-
-
-
- Gets the argument indexes for a node. Node can have multiple indexes, as opposed to GetArgumentIndex. Of course,
- calling this on PropBank nodes will only ever return a single index.
-
- Node to get argument indexes for. Must be a NomBankNode or a PropBankNode.
- Whether or not to convert NomBank argument indexes to PropBank indexes when
- possible
- Original argument indexes, before they were converted to PropBank indexes. If
- nomBankToPropBank is false, this must be null. If nomBankToPropBank is true, this may be either null or non-null. In
- the latter case, argument indexes will be added to the passed set using Set.Add. This set will also contain any PropBank
- argument indexes.
- Argument indexes
-
-
-
- Gets the predicate node for a given predicate tree
-
- Predicate tree - must be either a NomBankNode or a PropBankNode
- Whether or not to allow phrasal predicates. If true and a phrasal predicate is
- encountered, the first token of the phrasal predicate will be returned.
- Predicate node
-
-
-
- Gets base predicate for a predicate tree (i.e., the predicate contained in Information.Noun or Information.Verb). Only
- valid for root nodes.
-
- Predicate tree for which to get base predicate (must be root)
- Whether or not to convert NomBank predicates to PropBank predicates where possible
- Base predicate
-
-
-
- Gets base predicate for a predicate tree (i.e., the predicate contained in Information.Noun or Information.Verb). Only
- valid for root nodes.
-
- Predicate tree for which to get base predicate (must be root)
- Whether or not to convert NomBank predicates to PropBank predicates where possible
- Whether or not the returned predicate is a verb or a noun converted to a verb
- Base predicate
-
-
-
- Gets base predicate for a predicate tree (i.e., the predicate contained in Information.Noun or Information.Verb). Only
- valid for root nodes.
-
- Predicate tree for which to get base predicate (must be root)
- Whether or not to convert NomBank predicates to PropBank predicates where possible
- Whether or not the returned predicate is a verb or a noun converted to a verb
- Original predicate, before any NomBank-PropBank conversion
- Whether or not the original predicate is a verb
- Base predicate
-
-
-
- Gets argument nodes for a predicate tree. Only valid for root nodes of PropBank and NomBank trees.
-
- Predicate tree to get arguments for - must be a NomBankNode or PropBankNode
- Whether or not to include null-element argument nodes
- Whether or not to include split arguments
- If including split nodes, this specifies whether or not to only include the head node
- of the split argument. The head node is defined as the node containing the semantic head of the LCA of all nodes
- in the split argument.
- Whether or not to include single nodes
- Whether or not to exclude single nodes if there are more than one
- Set of argument nodes
-
-
-
- Gets the argument node collections for a TreeBankNode, which must be a NomBank or PropBank node.
-
- Node to get collections for
- Whether or not to remove null element nodes from the collections. This does not change the
- underlying PropBank or NomBank tree - null elements will still remain in these trees.
- Node collections
-
-
-
- Gets modifier nodes for a predicate tree. Only valid for root nodes of PropBank and NomBank trees.
-
- Predicate tree to get modifiers for - must be a NomBankNode or PropBankNode
- Whether or not to include null-element modifier nodes
- Whether or not to include split modifiers
- If including split nodes, this specifies whether or not to only include the head node
- of the split modifier. The head node is defined as the node containing the semantic head of the LCA of all nodes
- in the split modifier.
- Whether or not to include single nodes
- Whether or not to exclude single nodes if there are more than one
- Set of modifier nodes
-
-
-
- Tries to get information for a role in a predicate tree
-
- Predicate tree to get information
- Argument index to get information for
- Role description for the passed argument
- True if the role denoted by argumentIndex was found
-
-
-
- Tries to get information for a role in a predicate tree
-
- Predicate tree to get information
- Argument index to get information for
- Role description for the passed argument
- Source index for argument index
- True if the role denoted by argumentIndex was found
-
-
-
- Gets the role set for a TreeBankNode
-
- Node (must be root)
- Whether or not to convert NomBank role set IDs to PropBank role set IDs where possible
- Role set
-
-
-
- Gets unfilled roles for a PropBank or NomBank predicate tree
-
- Predicate tree to get unfilled roles for (must be PropBank or NomBank node)
- Whether or not to consider null-element nodes when checking whether a role is filled
- Unfilled roles
-
-
-
- Gets confidence of an argument node. Node must be either PropBankNode or NomBankNode, and must be an argument node.
-
- Node to get argument confidence for. Must be either PropBankNode or NomBankNode, and must be an argument node.
- Confidence of argument label
-
-
-
- Gets the WordNet POS for a TreeBank syntactic category
-
- TreeBank syntactic category to get WordNet POS for
- WordNet POS
-
Provides extension methods for the .NET Dictionary class
@@ -561,5 +212,189 @@
StreamReader to reset
Position to reset to
+
+
+ Provides extension methods for strings
+
+
+
+
+ Constructor
+
+
+
+
+ Replaces strings within a string
+
+ String to process
+ Replacement string pairs, where the key is the string to find and the value is the replacement
+ Whether or not to repeat replacement procedure until no changes are made
+ String with replacements made
+
+
+
+ Removes leading and trailing punctuation from a string
+
+ String to trim punctuation from
+ Trimmed string
+
+
+
+ Removes leading and trailing punctuation from a string
+
+ String to trim punctuation from
+ Whether or not to trim leading punctuation
+ Whether or not to trim trailing punctuation
+ Trimmed string
+
+
+
+ Removes punctuation characters (any that aren't a-z, A-Z, or 0-9)
+
+ String to process
+ String without punctuation
+
+
+
+ Replaces punctuation characters (any that aren't a-z, A-Z, or 0-9) with something else
+
+
+
+
+
+
+
+ Removes all whitespace characters from a string (\s regex character class)
+
+ String to process
+ String without whitespace
+
+
+
+ Removes repeated whitespace from a string
+
+ String to process
+ String without repeated whitespace
+
+
+
+ Throws an exception if any of the given characters are present in the string
+
+ String to check
+ Character(s) to disallow
+
+
+
+ Splits a string on space characters, guaranteeing a specific number of parts. Will throw an exception if the expected number of parts is not found.
+
+ String to split
+ Number of parts expected
+ Parts resulting from split
+
+
+
+ Splits a string on space characters, guaranteeing a specific number of parts. Will throw an exception if the expected number of parts is not found.
+
+ String to split
+ Number of parts expected
+ Characters to split on
+ Parts resulting from split
+
+
+
+ Gets enumeration of parts within a string, delimited by space characters
+
+
+
+
+
+
+ Converts a string to its XML-escaped version
+
+ Text to convert
+ XML-escaped version of text
+
+
+
+ Unescapes an string that has been XML-escaped
+
+ Text to convert
+ Unescaped XML text
+
+
+
+ Gets path relative to another path
+
+ Base path for absolute path
+ Absolute path
+ Relative path
+
+
+
+ Gets the common initial substring between two strings
+
+ First string
+ Second string
+ Common initial substring
+
+
+
+ Changes the first n characters of a string to uppercase
+
+ String to change
+ Number of characters to change
+ Modified string
+
+
+
+ Gets the index of the nth non-space character within a string
+
+ String to search
+ n
+ Index of nth non-space character
+
+
+
+ Gets the index of the nth occurrence of a character
+
+ String to search
+ Character to search for
+ Value of n
+ Index of the nth occurrence of c
+
+
+
+ Gets all indexes of a substring within the current string
+
+ String to search
+ Substring to search for
+ Indexes
+
+
+
+ Concatenates an enumeration of strings
+
+ Strings to concatenate
+ Separator for strings
+ Concatenated string
+
+
+
+ Encrypts a string using AES encryption
+
+ String to encrypt
+ Encryption key to use
+ Initialization to use
+ Encrypted bytes
+
+
+
+ Decrypts a string using AES encryption
+
+ Bytes to decrypt
+ Encryption key used to produce the bytes
+ Initialization that was used to produce the bytes
+ Unencrypted string
+
diff --git a/Libraries/LAIR.IO.dll b/Libraries/LAIR.IO.dll
index 6edf61f..fdcb4b1 100644
Binary files a/Libraries/LAIR.IO.dll and b/Libraries/LAIR.IO.dll differ
diff --git a/Libraries/LAIR.MachineLearning.dll b/Libraries/LAIR.MachineLearning.dll
index b290290..745de9c 100644
Binary files a/Libraries/LAIR.MachineLearning.dll and b/Libraries/LAIR.MachineLearning.dll differ
diff --git a/Libraries/LAIR.MachineLearning.xml b/Libraries/LAIR.MachineLearning.xml
index a57452e..f3a7736 100644
--- a/Libraries/LAIR.MachineLearning.xml
+++ b/Libraries/LAIR.MachineLearning.xml
@@ -4,20 +4,14 @@
LAIR.MachineLearning
-
-
- Logistic regression classifier
-
-
-
+
- Abstract base class for all Weka wrappers
+ Provides access to LibLinear classifier server
-
+
- Abstract base class for classifier wrappers that operate in a two-stage fashion, where each stage requires a call to an
- external executable program. These classifiers also make use of a directory to store various data.
+ Abstract base class for numbered feature classifier clients
@@ -130,1249 +124,1485 @@
Gets the file paths used to build models in this classifier
-
+
- Constructor
+ Constructor. Assumes a connection_params file is in model directory.
- Directory to store model files in
- Path to LibLinear learn executable
- Path to LibLinear classify executable
+ Method of access to use for feature name transform
+ Method of access to use for feature space
+ Whether or not to scale numeric feature values when classifying feature vectors
+ Model directory for this client. This directory must have a file named connection_params in it,
+ which has two lines. The first gives the host address, the second gives the host port.
+ Newline sequence used by classifier
Feature extractor to use
- Confidence threshold for labels
+ Label confidence threshold
-
+
Constructor
- Directory to store model files in
- Path to LibLinear learn executable
- Path to LibLinear classify executable
+ Method of access to use for feature name transform
+ Method of access to use for feature space
+ Whether or not to scale numeric feature values when classifying feature vectors
+ Model directory for this client. This directory must have a file named connection_params in it,
+ which has two lines. The first gives the host address, the second gives the host port.
+ Newline sequence used by classifier
Feature extractor to use
+ Label confidence threshold
+ Host for server
+ Port for server
-
+
- Gets argument list for invoked learn executable
+ Classifies a list of feature vectors at the remote server
- Argument list
+ Feature vectors to classify
-
+
- Learns classifier
+ Writes feature vectors to classify server
+ Feature vectors to be classified
-
+
- Cleans up
+ Reads predictions from server
+ Feature vectors that have been classified at server
-
+
- Gets argument list for invoked classify executable
+ Applies predictions to feature vectors
- Argument list
+ Feature vectors to apply predictions to
-
+
- Classifies a list of feature vectors
+ Gets textual representation of feature vector
- Feature vectors to classify
+ Feature vector to get text for
+ Feature vector text
-
+
- Cleans up
+ Gets mapped label for an unmapped label
-
+ Unmapped label to get mapped label for
+ Mapped label
-
+
- Gets or sets whether the learn process should be run after consuming all training vectors and firing the PreLearn event (default: true)
+ Not implemented
-
+
- Gets the learn executable path
+ Gets the name of the file that stores the connection parameters
-
+
- Gets the classify executable path
+ Gets the label for instances for which the class is unknown
-
+
- Gets the path to the executable's output
+ Gets the path to the feature name transform
-
+
- Gets or sets whether or not to throw an exception for any output on standard error when classifying with this classifier (default: true)
+ Gets the path to the feature space file
-
+
- Gets or sets whether or not to throw an exception for any output on standard error when learning the classifier (default: true)
+ Gets the path to the label map file
-
+
- Quote to use for values
+ Gets the label map, which maps application-level labels to classifier-level labels
-
+
- Matches the end quote in a quotes attribute value
+ Gets the path to the connection parameters file
-
+
- Gets the ARFF header for a list of features
+ Gets the path to the model file
- Features to get header for
- Class labels to write
- Attribute declaration
-
+
- Get instance text for a feature vector
+ Gets labels used by this classifier
- Feature vector to get instance text for
- Feature order to write
- Label map
- Instance text for feature vector
-
+
- Escapes an attribute value
+ Creates a LibLinear classification server startup script, and leaves LibLinear client connection parameter files in each directory
- Value to escape
- Escaped value
+ Parent directory, to be searched recursively for model files. The startup script is created in
+ this directory
+ File name for LibLinear models
+ LibLinear classify server host
+ Base port number for the LibLinear classify server
+ Whether or not to output probabilities if possible
-
+
+
+ Creates a LibLinear classification server startup script, and leaves LibLinear client connection parameters in each directory
+
+ Directory to be searched recursively for model files
+ File name for LibLinear models
+ LibLinear classify server host
+ Current port number for models
+ Path to the top-level directory
+ Path to top-level startup script
+ True if any entries were added to the startup script
+ Whether or not to output probabilities if possible
+
+
Constructor
+ Whether or not the server will output probabilities
+ Method of access to use for feature name transform
+ Method of access to use for feature space
+ Whether or not to scale numeric feature values when classifying feature vectors
+ Model directory for this client. This directory must have a file named liblinear_client_params in it,
+ which has two lines. The first gives the host address, the second gives the host port.
Feature extractor to use
- Model directory
- Path to Java executable file
- Path to Weka jar file
+ Label confidence threshold
-
+
- Clears files from model directory before consuming new training entities
+ Constructor
+ Whether or not the server will output probabilities
+ Method of access to use for feature name transform
+ Method of access to use for feature space
+ Whether or not to scale numeric feature values when classifying feature vectors
+ Model directory for this client. This directory must have a file named liblinear_client_params in it,
+ which has two lines. The first gives the host address, the second gives the host port.
+ Feature extractor to use
+ Label confidence threshold
+ Host for server
+ Port for server
-
+
- Consumes training vectors
+ Applies current predictions to a list of feature vectors
- Vectors to consume
+ Feature vectors to apply predictions to
-
+
- Finishes ARFF file
+ Gets the label for instances for which the true class is unknown
-
+
- GetLearnProcessArguments override
+ Gets or sets whether or not the server will output probabilities
-
-
+
- Cleans up after learn process runs
+ Reads LibLinear-style feature vector files
-
+
- Gets ready to classify feature vectors
+ Reads FeatureVectorStrings from a text file
- Feature vectors to be classified
-
+
- Gets classify process's arguments
+ Gets feature vector strings from a file
-
+ Path to file
+ List of vectors
-
+
- Applies predictions to feature vectors
+ Gets feature vector strings from a file
- FeatureVectors that were classified
+ Path to file from which to get vectors
+ List of feature vector strings
-
+
- Gets labels used by this classifier
+ Extracts features from FeatureVectorStrings that are in LibLinear format
-
+
- Multi-class strategy to use (default: None)
+ Abstract base class for classes that extract feature vectors from classifiable entities
-
+
- Whether or not to use the server JVM instead of the client JVM (default: false)
+ Constructor
+ Whether or not to use abbreviated feature names
+ Whether or not to use abbreviated feature values
-
+
- Gets or sets whether or not to perform cross validation (default: false)
+ Gets a feature vector for a single entity
+ Entity to get feature vector for
+ Feature vector
-
+
- Gets or sets whether or not to delete the training file after it is used (default: false)
+ Gets a feature vector for a single entity
+ Entity to get feature vector for
+ Estimated number of features that might be extracted for the entity. Passing this argument
+ allows feature extraction to initialize the feature vector with a predetermined capacity, which is good for memory usage.
+ Feature vector
-
+
- Gets command line arguments for the JVM
+ Gets feature vectors for a list of entities
+ Entities to process
+ Number of features that might be extracted for an entity
+ List of feature vectors
-
+
- Gets multi-class classifier arguments
+ Gets feature vectors for a list of entities
+ Entities to process
+ List of feature vectors
-
+
- Gets the class for multi-class classification
+ Gets or sets whether or not to use abbreviated feature values instead of full-length values
-
+
- Gets the classifier class
+ Gets or sets whether or not to use abbreviated feature names instead of the full names
-
+
- Gets the path to the training instances file
+ Constructor
-
+
- Get the path to the classification instances file
+ Gets a FeatureVector from a FeatureVectorString object
+ Entity to get feature vector from
+ Features per entity
+ FeatureVector
-
+
- Gets the file to save label map to
+ Wrapper for the LibLinear classifier
-
+
+
+ Abstract base class for all classifiers that make use of the following instance format:
+
+ label feature1:value1 feature2:value2 ... featureN:valueN
+
+ The one-based feature numbers are sorted ascendingly from left to right. Example classifiers
+ include SVM Light (and its derivatives) and LibLinear.
+
+
+
- Gets the feature name transform save file
+ Abstract base class for classifier wrappers that operate in a two-stage fashion, where each stage requires a call to an
+ external executable program. These classifiers also make use of a directory to store various data.
-
+
- Gets the path to the feature space file
+ Constructor
+ Directory to store model files in
+ Path to LibLinear learn executable
+ Path to LibLinear classify executable
+ Feature extractor to use
+ Confidence threshold for labels
-
+
- Gets the path to the feature order file
+ Constructor
+ Directory to store model files in
+ Path to LibLinear learn executable
+ Path to LibLinear classify executable
+ Feature extractor to use
-
+
- Gets the model file
+ Gets argument list for invoked learn executable
+ Argument list
-
+
- Gets list of file used in the curren tmodel
+ Learns classifier
-
+
- Types of multi-class strategies
+ Cleans up
-
+
- Constructor
+ Gets argument list for invoked classify executable
- Feature extractor
- Model directory
- Path to Java executable
- Path to Weka jar file
+ Argument list
-
+
- Gets the logistic regression class
+ Classifies a list of feature vectors
+ Feature vectors to classify
-
+
- Represents a feature vector used by the Cluto toolkit
+ Cleans up
+
-
+
- Represents a sparse vector of features
+ Gets or sets whether the learn process should be run after consuming all training vectors and firing the PreLearn event (default: true)
-
+
- Constructor
+ Gets the learn executable path
- Entity this vector is derived from
-
+
+
+ Gets the classify executable path
+
+
+
+
+ Gets the path to the executable's output
+
+
+
+
+ Gets or sets whether or not to throw an exception for any output on standard error when classifying with this classifier (default: true)
+
+
+
+
+ Gets or sets whether or not to throw an exception for any output on standard error when learning the classifier (default: true)
+
+
+
+
+ Gets instance text for feature vector
+
+ Feature vector to get instance text for
+ Label map to use
+ Gets the label for instances for which the true class is unknown
+ Feature space to restrict feature space to, or null for no restriction
+ Whether or not to scale numeric feature values to [0,1] using the observed range in
+ the given feature space. It's possible for a scaled value to be greater than 1 in cases where the observed range doesn't
+ include the value.
+ Feature name transform to apply
+ Instance text for a feature vector
+
+
+
+ Gets instance text for feature vector
+
+ Feature vector to get instance text for
+ Label map to use
+ Gets the label for instances for which the true class is unknown
+ Feature space to restrict feature space to, or null for no restriction
+ Whether or not to scale numeric feature values to [0,1] using the observed range in
+ the given feature space. It's possible for a scaled value to be greater than 1 in cases where the observed range doesn't
+ include the value. If true, caller must supply a feature space.
+ Feature name transform to apply
+ This method will add any numeric feature numbers to the supplied set. Pass null if
+ feature numbers aren't needed.
+ Instance text for a feature vector
+
+
+
+ Changes the names of features used by this classifier
+
+ Directory of model
+ Old names and new names
+
+
Constructor
- Entity this vector is derived from
- Number of features this vector will contain
+ Method of access for feature name transform
+ Method of access for feature space
+ Label mapping to perform
+ Scale numeric feature values
+ Directory to store model files in
+ Newline sequence to use for classifier
+ Path to LibLinear learn executable
+ Path to LibLinear classify executable
+ Feature extractor to use
+ Confidence threshold for labels
-
+
- Adds a feature/value pair to this vector
+ Gets an unmapped label
- Feature to add
- Value of feature to add
- Whether or not to update the stored range on the added feature
+ Mapped label to get unmapped label for
+ Unmapped label
-
+
- Adds feature/value pairs from a vector to this vector
+ Gets a mapped label
- Vector whose feature/value pairs should be added to this one
+ Unmapped label to get mapped label for
+ Mapped label
-
+
- Removes a feature from this vector
+ Gets 1-based feature number given a feature name
- Feature to remove
+ Name of feature to transform
+ Feature number
-
+
- Removes a feature from this vector
+ Tries to get the 1-based feature number given a feature name
- Name of feature to remove
+ Name of feature to transform
+ Feature number (output)
+ True if feature number was retrieved and false otherwise
-
+
- Gets whether or not this feature vector contains a feature
+ Gets instance text as it would appear when classifying a feature vector
- Feature to check
- True if feature is present, false otherwise
+ Feature vector to get instance text for
+ Instance text
-
+
- Gets whether or not this feature vector contains a feature by name
+ Called just before the first training vectors are consumed
- Feature name to check
- True if feature is present, false otherwise
-
+
- Tries to get a feature value from this vector
+ Consumes training feature vectors
- Feature to get value for
- Value of feature
- True if value was retrieved, false otherwise
+
-
+
- Tries to get a feature value from this vector by name
+ Called just before learning takes place
- Feature name to get value for
- Value of feature
- True if value was retrieved, false otherwise
-
+
- Clears all feature/value pairs from this vector
+ Called after learning takes place
-
+
- Normalizes the current vector to be of unit length. This only works if all feature values can be casted into floats.
+ Gets the index of the space character that comes before the feature-value pairs
+ Feature vector line
+ Index of space
-
+
- Gets printable string value of this vector
+ Write instance file to classify
-
+ Feature vectors to be classified
-
+
- Restricts the feature space of this vector to conform with the given feature space. Features in this vector that are
- not contained in the feature space will be removed. For restrictions on the feature values, see the parameters.
+ Called just after classification takes place
- Feature space to use
- Whether or not to require nominal features in this vector to have values that are
- present in the range of the corresponding feature in the given feature space.
- Whether or not to require numeric features in this vector to have values that are
- within the range of the corresponding feature in the given feature space.
+ Feature vectors that were classified
-
+
- Gets enumerator over features in this vector
+ Loads supporting model files that allow this classifier to classify feature vectors
- Enumerator over features in this vector
-
+
- Gets enumerator over features in this vector
+ Gets the name of the training instances file
- Enumerator over features in this vector
-
+
- Gets the Euclidean length of this vector. All feature values must be castable to floats.
+ Gets the feature name transform file name
-
+
- Gets the value of a given feature
+ Gets the feature space file name
- Feature to get value for
- Value of feature
-
+
- Gets the value of a given feature
+ Gets the label map file name
- Name of feature
- Value of feature
-
+
- Gets number of features in this vector
+ Gets the model file name
-
+
- Gets the collection of features used in this vector
+ Gets the name of the instances to classify file
-
+
- Gets or sets the entity from which this feature vector is derived
+ Gets the label for instances for which the class is unknown
-
+
+
+ Gets the newline sequence used by this classifier
+
+
+
+
+ Gets the name transform access method
+
+
+
+
+ Gets the feature space access method
+
+
+
+
+ Gets or sets the feature space condensation method (default: DiskBased)
+
+
+
+
+ Gets whether or not to scale numeric feature values
+
+
+
+
+ Gets the path to the file containing training instances
+
+
+
+
+ Gets the path to the feature name transform
+
+
+
+
+ Gets the path to the feature space file
+
+
+
+
+ Gets the path to the feature space range search file
+
+
+
+
+ Gets the path to the label map
+
+
+
+
+ Gets the label map used in this classifier
+
+
+
+
+ Gets labels used by this classifier
+
+
+
+
+ Gets the path to the model
+
+
+
+
+ Gets the path to the file containing the instances to classify
+
+
+
+
+ Gets the model files used by the implementing class
+
+
+
+
+ Gets or sets the starting size of the memory-based feature name transform. In order to take effect, this must
+ be set before the first training vectors are consumed. Will throw an exception if set when using disk-based
+ feature name transformation.
+
+
+
+
+ Gets or sets whether to delete the training instances file after learning has completed (default: false)
+
+
+
+
+ Gets whether a solver is probabilistic
+
+
+
+
+
+
+ Gets the solver type for a the model ID given in the model file
+
+ ID from model file
+ Solver type
+
+
+
+ Applies classification predictions to a list of feature vectors
+
+ Path to predictions file
+ Feature vectors to apply predictions to
+ Whether or not probabilities were output
+ Label map to use
+
+
+
+ Gets feature weights from a LibLinear model file. First key is the class label, the second key is the 1-based feature number
+ and the value is the weight. For a two-class problem, there is only one class label in the first key, and that is the label
+ first encountered in the model file. For a multi-class problem, there is a key for each class label.
+
+ Path to LibLinear model file
+ Feature weights
+
+
+
+ Creates a CSV file from a LibLinear model
+
+
+
+
+
+
+ Quotes a string if needed for use in CSV
+
+
+
+
+
Constructor
- Entity this vector is derived from
- Internal z-score
- External z-score
+ Solver to use
+ Method of access for feature name transform
+ Method of access for feature space
+ Scale numeric feature values
+ Directory to store model files in
+ Path to LibLinear learn executable
+ Path to LibLinear classify executable
+ Feature extractor to use
+ Confidence threshold for labels
-
+
- Gets or sets the internal z-score for this vector
+ Gets learn process arguments
+ Learn process arguments
-
+
- Gets or sets the external z-score for this vector
+ Gets the index of the space character that comes before the feature-value pairs
+ Feature vector
+ Index of space
-
+
+
+ Gets arguments for learn process
+
+ Arguments for learn process
+
+
+
+ Applies predictions and cleans up
+
+ Feature vectors that have been classified
+
+
+
+ Gets the name of the predictions file name
+
+
+
- Compares two feature vectors using the values of their internal z-scores
+ Gets the label for instances for which the true class is unknown
-
+
- Compares two feature vectors using the values of their internal z-scores
+ Gets the type of solver used
- First vector
- Second vector
-
-
+
- Compares two feature vectors using the values of their external z-scores
+ Gets or sets the cost parameter
-
+
- Compares two feature vectors using the values of their external z-scores
+ Gets or sets the epsilon parameter
- First vector
- Second vector
-
-
+
- Compares two feature vectors using a combination of the internal and external z-scores
+ Gets or sets the bias parameter
-
+
- Compares two feature vectors using a combination of the internal and external z-scores
+ Gets or sets the number of cross-validation folds to use
- First vector
- Second vector
-
-
+
- Provides instances to classifiers, annotators, etc. - anything that needs a stream of classifiable instances
+ Gets or sets whether or not to output classification probabilities (only valid for logistic regression classifier)
-
+
- Constructor
+ Gets the path to the file to be written with predictions
-
+
- Starts the instance stream
+ Gets a list of files used in the current model
-
+
- Gets the next instance in the stream
+ Gets or sets the per-class weights
- Instance
-
+
- Gets the previous instance in the stream
+ Types of LibLinear solvers
- Instance
-
+
- Gets the zero-based number of the current instance
+ L2-regularized primal logistic regression (s=0)
-
+
- Gets the total number of instances available
+ L2-regularized L2-loss dual SVM (s=1)
-
+
- Gets the number of instance remaining
+ L2-regularized L2-loss primal SVM (s=2)
-
+
- Gets whether or not there is another instance available
+ L2-regularized L1-loss dual SVM (s=3)
-
+
- Gets whether or not there is a previous instance
+ Multi-class SVM by Crammer and Singer (s=4)
-
+
- Gets the current training instance
+ L1-regularized L2-loss SVM (s=5)
-
+
- Reads SvmLight-style feature vector text files
+ L1-regularized logistic regression (s=6)
-
+
- Reads FeatureVectorStrings from a text file
+ L2-regularized dual logistic regression (s=7)
-
+
- Gets feature vector strings from a file
+ Reads SvmRank-style feature vector text files
- Path to file
- List of vectors
-
+
Gets feature vectors from a file
Path to file from which to get vectors
List of feature vectors
-
+
- Extracts features from FeatureVectorStrings that are in LibLinear format
+ Wrapper for the SVM Rank classifier. Any calls to classify assume that the classified feature
+ vectors are to be ranked against each other.
-
+
- Abstract base class for classes that extract feature vectors from classifiable entities
+ Applies predictions from a file to classified feature vectors
+ Prediction output
+ Feature vectors to apply predictions to
-
+
Constructor
- Whether or not to use abbreviated feature names
- Whether or not to use abbreviated feature values
-
-
-
- Gets a feature vector for a single entity
-
- Entity to get feature vector for
- Feature vector
+ Trade-off between training error and margin
+ Method of access for feature name transform
+ Method of access for feature space
+ Scale numeric feature values
+ Directory to store model files in
+ Path to LibLinear learn executable
+ Path to LibLinear classify executable
+ Feature extractor to use
-
+
- Gets a feature vector for a single entity
+ Gets the index of the space character that comes before the feature-value pairs
- Entity to get feature vector for
- Estimated number of features that might be extracted for the entity. Passing this argument
- allows feature extraction to initialize the feature vector with a predetermined capacity, which is good for memory usage.
- Feature vector
+ Feature vector
+ Index of space
-
+
- Gets feature vectors for a list of entities
+ Gets learning process arguments
- Entities to process
- Number of features that might be extracted for an entity
- List of feature vectors
+
-
+
- Gets feature vectors for a list of entities
+ Gets classification process arguments
- Entities to process
- List of feature vectors
+
-
+
- Gets or sets whether or not to use abbreviated feature values instead of full-length values
+ Called just before classification. Sets the qid for each feature vector to be the same.
+ Feature vectors to rank
-
+
- Gets or sets whether or not to use abbreviated feature names instead of the full names
+ Applies predictions and cleans up
+ Feature vectors that have been classified
-
+
- Constructor
+ Gets the name of the predictions file name
-
+
- Gets a FeatureVector from a FeatureVectorString object
+ Gets the label for instances for which the true class is unknown
- Entity to get feature vector from
- Features per entity
- FeatureVector
-
+
- Provides access to LibLinear classifier server
+ Gets the path to the predictions file
-
+
- Abstract base class for numbered feature classifier clients
+ Gets files used in SVM Light models
-
+
- Constructor. Assumes a connection_params file is in model directory.
+ Represents a feature in Weka
- Method of access to use for feature name transform
- Method of access to use for feature space
- Whether or not to scale numeric feature values when classifying feature vectors
- Model directory for this client. This directory must have a file named connection_params in it,
- which has two lines. The first gives the host address, the second gives the host port.
- Newline sequence used by classifier
- Feature extractor to use
- Label confidence threshold
-
+
- Constructor
+ Gets the value to use when an instance is missing a feature
- Method of access to use for feature name transform
- Method of access to use for feature space
- Whether or not to scale numeric feature values when classifying feature vectors
- Model directory for this client. This directory must have a file named connection_params in it,
- which has two lines. The first gives the host address, the second gives the host port.
- Newline sequence used by classifier
- Feature extractor to use
- Label confidence threshold
- Host for server
- Port for server
-
+
- Classifies a list of feature vectors at the remote server
+ Extracts features from FeatureVectorStrings that are in LibLinear format
- Feature vectors to classify
-
+
- Writes feature vectors to classify server
+ Constructor
- Feature vectors to be classified
-
+
- Reads predictions from server
+ Gets a FeatureVector from a FeatureVectorString object
- Feature vectors that have been classified at server
+ Entity to get feature vector from
+ Features per entity
+ FeatureVector
-
+
- Applies predictions to feature vectors
+ Represents a nominal feature in Weka
- Feature vectors to apply predictions to
-
+
- Gets textual representation of feature vector
+ Represents non-numeric features
- Feature vector to get text for
- Feature vector text
-
+
- Gets mapped label for an unmapped label
+ Abstract base class for all feature classes
- Unmapped label to get mapped label for
- Mapped label
-
+
- Not implemented
+ Constructor
+ Name of feature. May not contain spaces.
-
+
- Gets the name of the file that stores the connection parameters
+ Updates the range of this feature
+
-
+
- Gets the label for instances for which the class is unknown
+ Checks whether this feature equals another
+ Other feature to check
+ True if features are equal, false otherwise
-
+
- Gets the path to the feature name transform
+ Gets hash code for this feature
+ Hash code
-
+
- Gets the path to the feature space file
+ Gets a deep copy of this feature
-
+
- Gets the path to the label map file
+ Renames the current feature, returning a new feature.
+ New feature name
+ New feature
-
+
- Gets the label map, which maps application-level labels to classifier-level labels
+ Gets name of feature
+
-
+
- Gets the path to the connection parameters file
+ Compares this feature to another feature on the basis of feature names
+ Other feature
+
-
+
- Gets the path to the model file
+ Gets or sets the name
-
+
- Gets labels used by this classifier
+ Constructor
+ Name of feature
-
+
- Creates a LibLinear classification server startup script, and leaves LibLinear client connection parameter files in each directory
+ Constructor
- Parent directory, to be searched recursively for model files. The startup script is created in
- this directory
- File name for LibLinear models
- LibLinear classify server host
- Base port number for the LibLinear classify server
- Whether or not to output probabilities if possible
+ Name of feature
+ Initial range size
-
+
- Creates a LibLinear classification server startup script, and leaves LibLinear client connection parameters in each directory
+ Constructor
- Directory to be searched recursively for model files
- File name for LibLinear models
- LibLinear classify server host
- Current port number for models
- Path to the top-level directory
- Path to top-level startup script
- True if any entries were added to the startup script
- Whether or not to output probabilities if possible
+ Name of feature
+ Elements to initialize range with
-
+
- Constructor
+ Updates the range of this nominal feature
- Whether or not the server will output probabilities
- Method of access to use for feature name transform
- Method of access to use for feature space
- Whether or not to scale numeric feature values when classifying feature vectors
- Model directory for this client. This directory must have a file named liblinear_client_params in it,
- which has two lines. The first gives the host address, the second gives the host port.
- Feature extractor to use
- Label confidence threshold
+
-
+
- Constructor
+ Checks whether or not a value is in the range of this feature
- Whether or not the server will output probabilities
- Method of access to use for feature name transform
- Method of access to use for feature space
- Whether or not to scale numeric feature values when classifying feature vectors
- Model directory for this client. This directory must have a file named liblinear_client_params in it,
- which has two lines. The first gives the host address, the second gives the host port.
- Feature extractor to use
- Label confidence threshold
- Host for server
- Port for server
+ Value to check
+ Try if value is in the range, false otherwise
-
+
- Applies current predictions to a list of feature vectors
+ Adds a value to this feature's range, or does nothing if the value already exists in the range
- Feature vectors to apply predictions to
+ Value to add
+ True if value was added as new, false if it was already in the range
-
+
- Gets the label for instances for which the true class is unknown
+ Gets a deep copy of this nominal feature
+ NominalFeature
-
+
- Gets or sets whether or not the server will output probabilities
+ Renames the current feature, returning a new feature.
+ New name
+ New feature
-
+
- Sorts features by name
+ Gets whether or not this nominal feature equals another feature
+ Other feature
+ True if both features are nominal with the same name
-
+
- Compares two features by name
+ Gets hash code for this feature
- First feature
- Second feature
- -1 if x comes before y, 1 if x comes after y, and 0 otherwise
+ Hash code
-
+
- Extracts features from FeatureVectorStrings that are in LibLinear format
+ Gets the range of values for this nominal feature
-
+
Constructor
+
-
+
- Gets a FeatureVector from a FeatureVectorString object
+ Constructor
- Entity to get feature vector from
- Features per entity
- FeatureVector
+
+
-
+
- Maps class labels to a form usable by a classifier
+ Constructor
+
+
+
-
+
- Constructor. The resulting LabelMap will be unlocked.
+ Constructor
- Type of mapping to perform
+
+
+
-
+
- Constructor
+ Gets copy
- Path to saved label map
+
-
+
- Adds a mapping
+ Renames
- Unmapped label
- Mapped label
+
+
-
+
- Gets the mapped label for an unmapped label
+ Gets value to use when feature is missing
- Label to get mapped label for
- Mapped label
-
+
- Gets unmapped label for a mapped label
+ Represents a numeric feature in Weka
- Mapped label to get unmapped label for
- Unmapped label
-
+
- Saves this map to disk
+ Represents features with infinite numeric domains
- Where to write map to
-
+
- Clears this label map
+ Constructor
+ Name of feature
-
+
- Checks whether this map contains a mapped label for a given unmapped label
+ Constructor
- Label to check for
- True if such a label exists, false otherwise
+ Name of feature
+ Minimum value of this feature
+ Maximum value of this feature
-
+
- Gets or sets whether or not this LabelMap is locked (default: false)
+ Updates the range of this numeric feature
+
-
+
- Gets unmapped labels, which are used by the application
+ Gets a deep copy of this numeric feature
+ Feature copy
-
+
- Gets mapped labels, which are used by the classifier
+ Renames the current feature, returning a new feature.
+ New name
+ New feature
-
+
- Mapping types
+ Gets whether or not this numeric feature equals another feature
+ Other feature
+ True if both features are numeric with the same name
-
+
- Perform no mapping. Directly use the given label.
+ Gets hash code for this feature
+ Hash code
-
+
- Map labels to 1-based increasing integers
+ Gets whether or not a value is within the range of this numeric feature
+ Value to check
+ True if the value is in the range, false otherwise
-
+
- Represents a feature space that can be searched using a binary search over a file. This is a read-only feature space.
+ Updates the range on this feature
+ Minimum value
+ Maximum value
-
+
- Provides disk-based access to a feature space. Creates an auxiliary file that maps feature names to their ranges - numerical
- ranges in the case of numerical features, and byte ranges in the case of nominals features, where the byte range specifies
- a portion of the feature space file to search for the feature value.
+ Gets or sets the minimum value of this feature
-
+
- Represents a feature spaces that FeatureVectors reside in
+ Gets or sets the maximum value of this feature
-
+
- Condenses a feature space file. When multiple feature spaces are appended to the same file, the same feature can be listed
- multiple times. This results in redundancies that must be removed for proper operation over the feature space file.
+ Constructor
- Feature space file to condense
- Output file for condensed feature space (may be the same as the input path)
- Method of condensing the feature space
+
+
-
+
Constructor
+
+
+
+
-
+
- Clears this feature space and releases any resources held by it
+ Gets copy
+
-
+
- Gets whether or not this space contains a feature
+ Renames
+
+
-
+
- Gets whether or not this space contains a feature/value pair
+ Gets value to use when feature is missing
- Feature to check
- Value to check
- True if feature/value pair is present, false otherwise
-
+
- Gets the range of a numeric feature value
+ Represents a discriminative Markov model
- Numeric feature to get range for
- Minimum value of range
- Maximum value of range
-
+
- Scales a numeric feature value to be within [-1,1] based on the observed range of the feature in the training data. This
- isn't guaranteed, as the given feature value might be outside the feature's observed range range defined in this space, in
- which case the value might be outside of [-1,1]. The range of [-1,1] is usually appropriate, since many classifiers will
- assume a value of 0 for missing features. The value 0 in the range [-1,1] is uninformative for any class and thus
- appropriate for features missing a value.
+ Represents
- Numeric feature to scale
- Value to scale
- Scaled value
-
+
- Gets or sets whether or not this feature space is locked (default: false)
+ Constructor
+
+
-
+
- Gets number of features in this space
+ Creates a state
+
+
+
-
+
- Access methods
+ Gets states at a time index
+ Time at which to get states
+ States
-
+
- In-memory access
+ Gets the observation at a given time
+ Observation at the given time
-
+
- Hash search-based access
+ Runs the model, calculating Viterbi and forward probabilities.
-
+
- Binary search-based access
+ Gets the forward probability of a state at a time, given a previous state.
+
+
+
+
-
+
- Methods for condensing feature spaces
+ Gets the Viterbi probability of a state at a time, given a previous state.
+
+
+
+
-
+
- Uses little memory, but creates large numbers of files on disk
+ Decodes the most likely state sequence
+
-
+
- Creates no auxiliary files, but uses lots of memory. Only recommended for small feature spaces.
+ Gets observations
-
+
- Constructor
+ Gets the start states
-
+
- Creates an auxiliary file that maps feature names to their ranges - numerical ranges in the case of numerical features,
- and byte ranges in the case of nominals features, where the byte range specifies a portion of the feature space file to
- search for the feature value.
+ Gets the end states
- Path to feature space for which to create range file
- Whether or not to create a range file in which the feature names are sorted. This is required,
- for example, in the case of binary searches over ordered data
- Path to range file that was created
-
+
- Gets whether or not this space contains a feature
+ Gets the most likely end state
- Feature to check for
- True if feature is contained, false otherwise
-
+
- Gets whether or not this feature space contains a feature/value pair
+ Gets number of time steps in the model
- Feature to check
- Value of feature
- True if feature/value pair is not present
-
+
- Gets the search value prefix for a feature-value query
+ Gets the number of states in each time step
- Feature to get prefix for
- Search value prefix for a feature-value query
-
+
- Closes the feature space files
+ Gets the state IDs
-
+
- Gets the range of a numeric feature value
+ Constructor
- Numeric feature to get range for
- Minimum value of range
- Maximum value of range
+
+
+
-
+
- Gets or sets the feature space search
+ Creates a discriminative state
+
+
+
-
+
- Gets or sets the feature range search
+ Gets the forward probability of a state
+
+
+
+
-
+
- Gets the number of features in this feature space
+ Gets the Viterbi probability
+
+
+
+
-
+
- Gets whether or not the feature space is locked (always true, cannot be set to false)
+ Classifies all times at once
-
+
- Constructor
+ Provides training instances for a discriminative HMM
- Path to feature space file that is to be searched
-
+
- The binary search feature space does not need to do feature-value prefixing because it does not do any tricky
- indexing of feature values. It simply conducts a binary search for a feature value within a byte range. This
- function always returns the empty string.
+ Provides training instances
- Feature to get prefix for
- Empty string
-
+
- Transforms string feature names to numeric feature names. Some classification frameworks only work with numeric features names,
- so we must have a method of transforming string names to numeric names.
+ Provides instances to classifiers, annotators, etc. - anything that needs a stream of classifiable instances
-
+
Constructor
- Whether or not transform is locked
-
-
-
- Gets 1-based feature number given a feature name
-
- Name of feature to transform
- Feature number
-
+
- Gets the feature name for a feature number
+ Starts the instance stream
- Feature number to get name for
- Feature name
-
+
- Tries to get the feature name for a feature number
+ Gets the next instance in the stream
- Feature number to get name for
- Feature name
- Feature name
+ Instance
-
+
- Tries to get the 1-based feature number given a feature name
+ Gets the previous instance in the stream
- Name of feature to transform
- Feature number (output)
- True if feature number was retrieved and false otherwise
+ Instance
-
+
- Closes this transform, releasing resources
+ Gets the zero-based number of the current instance
-
+
- Saves this transform to disk
+ Gets the total number of instances available
-
+
- Gets or sets whether or not this transform is locked (default: false)
+ Gets the number of instance remaining
-
+
- Available acess methods
+ Gets whether or not there is another instance available
-
+
- All data is stored in memory
+ Gets whether or not there is a previous instance
-
+
- Feature name hashes are stored in memory, but numeric values are on disk
+ Gets the current training instance
-
+