BordeauxClassifierpublic class BordeauxClassifier extends Object | Classifier for the Learning framework.
For training: call trainOneSample
For classifying: call classify
Data is represented as sparse key, value pair. And key is an integer, value
is a float. Class label(target) for the training data is an integer.
Note: since the actual classifier is running in a remote the service.
Sometimes the connection may be lost or not established. |
| Fields Summary |
|---|
| static final String | TAG | | private android.content.Context | mContext | | private String | mName | | private android.bordeaux.services.ILearning_MulticlassPA | mClassifier |
| Constructors Summary |
|---|
public BordeauxClassifier(android.content.Context context)
mContext = context;
mName = "defaultClassifier";
mClassifier = BordeauxManagerService.getClassifier(context, mName);
| public BordeauxClassifier(android.content.Context context, String name)
mContext = context;
mName = name;
mClassifier = BordeauxManagerService.getClassifier(context, mName);
|
| Methods Summary |
|---|
| public int | classify(java.util.HashMap sample)
// if classifier is not available return -1 as an indication of fail.
if (!retrieveClassifier())
return -1;
try {
return mClassifier.Classify(getArrayList(sample));
} catch (RemoteException e) {
Log.e(TAG,"Exception: classify the sample.");
// return an invalid number.
// TODO: throw exception.
return -1;
}
| | private java.util.ArrayList | getArrayList(java.util.HashMap sample)
ArrayList<IntFloat> intfloat_sample = new ArrayList<IntFloat>();
for (Map.Entry<Integer, Float> x : sample.entrySet()) {
IntFloat v = new IntFloat();
v.index = x.getKey();
v.value = x.getValue();
intfloat_sample.add(v);
}
return intfloat_sample;
| | private boolean | retrieveClassifier()
if (mClassifier == null)
mClassifier = BordeauxManagerService.getClassifier(mContext, mName);
// if classifier is not available, return false
if (mClassifier == null) {
Log.i(TAG,"Classifier not available.");
return false;
}
return true;
| | public boolean | update(java.util.HashMap sample, int target)
if (!retrieveClassifier())
return false;
try {
mClassifier.TrainOneSample(getArrayList(sample), target);
} catch (RemoteException e) {
Log.e(TAG,"Exception: training one sample.");
return false;
}
return true;
|
|
