Package org.tribuo.hash

Class HashedFeatureMap

java.lang.Object

org.tribuo.FeatureMap

org.tribuo.ImmutableFeatureMap

org.tribuo.hash.HashedFeatureMap

All Implemented Interfaces:

Serializable, Iterable<VariableInfo>, ProtoSerializable<org.tribuo.protos.core.FeatureDomainProto>

public final class HashedFeatureMap
extends ImmutableFeatureMap

A FeatureMap used by the HashingTrainer to provide feature name hashing and guarantee that the Model does not contain feature name information, but still works with unhashed features names.

The salt must be set after this object has been deserialized.

See Also:

• Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
static final int	CURRENT_VERSION	Protobuf serialization version.

Fields inherited from class org.tribuo.ImmutableFeatureMap

idMap, size

Fields inherited from class org.tribuo.FeatureMap

m

Fields inherited from interface org.tribuo.protos.ProtoSerializable

DESERIALIZATION_METHOD_NAME, PROVENANCE_SERIALIZER

Method Summary

Modifier and Type	Method	Description
static HashedFeatureMap	deserializeFromProto(int version, String className, com.google.protobuf.Any message)	Deserialization factory.
static HashedFeatureMap	generateHashedFeatureMap(FeatureMap map, Hasher hasher)	Converts a standard FeatureMap by hashing each entry using the supplied hash function Hasher.
VariableIDInfo	get(String name)	Gets the VariableIDInfo for this name.
int	getID(String name)	Gets the id number for this feature, returns -1 if it's unknown.
void	setSalt(String salt)	The salt is not serialised with the Model.

Methods inherited from class org.tribuo.ImmutableFeatureMap

generateIDs, generateIDs, get, serialize, size

Methods inherited from class org.tribuo.FeatureMap

deserialize, domainEquals, equals, hashCode, iterator, keySet, toReadableString, toString

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface java.lang.Iterable

forEach, spliterator

Field Details

CURRENT_VERSION

public static final int CURRENT_VERSION

Protobuf serialization version.

See Also:

• Constant Field Values

Method Details

deserializeFromProto

public static HashedFeatureMap deserializeFromProto(int version,
 String className,
 com.google.protobuf.Any message)
                                             throws com.google.protobuf.InvalidProtocolBufferException

Deserialization factory.

Parameters:

version - The serialized object version.

className - The class name.

message - The serialized data.

Returns:

The deserialized object.

Throws:

com.google.protobuf.InvalidProtocolBufferException - If the protobuf could not be parsed from the message.

get

public VariableIDInfo get(String name)

Description copied from class: ImmutableFeatureMap

Gets the VariableIDInfo for this name. Returns null if it's unknown.

Overrides:

get in class ImmutableFeatureMap

Parameters:

name - The name to lookup.

Returns:

The VariableInfo, or null.

getID

public int getID(String name)

Gets the id number for this feature, returns -1 if it's unknown.

Overrides:

getID in class ImmutableFeatureMap

Parameters:

name - The name of the feature.

Returns:

A non-negative integer if the feature is known, -1 otherwise.

setSalt

public void setSalt(String salt)

The salt is not serialised with the Model. It must be set after deserialisation to the same value from training time.

If the salt is invalid it will throw IllegalArgumentException.

Parameters:

salt - The salt value. Must be the same as the one from training time.

generateHashedFeatureMap

public static HashedFeatureMap generateHashedFeatureMap(FeatureMap map,
 Hasher hasher)

Converts a standard FeatureMap by hashing each entry using the supplied hash function Hasher.

This preserves the index ordering of the original feature names, which is important for making sure test time performance is good.

It guarantees any collisions will produce a feature id number lower than the previous feature's number, and so can be easily removed.

Parameters:

map - The FeatureMap to hash.

hasher - The hashing function.

Returns:

A HashedFeatureMap.