Interaction Model¶

InteractionModel¶

class ml4ir.base.model.scoring.interaction_model.InteractionModel¶

Bases: object

InteractionModel class that defines tensorflow layers that act on input features to convert them into numeric features to be fed into further neural network layers

feature_layer_op(inputs: Dict[str, tensorflow.python.keras.engine.input_layer.Input])¶

Convert input tensorflow features into numeric train features and metadata features by applying respective feature transformation functions as specified in the FeatureConfig

Parameters:	inputs (dict) – Dictionary of the inputs to the tensorflow keras model
Returns:	train_features (dict) – Dict of feature tensors that are used for training metadata_features (dict) – Dictionary of feature tensors that can be used for computing custom metrics and losses

transform_features_op(train_features, metadata_features)¶

Convert train and metadata features which have feature layer functions applied to them into dense numeric tensors

Parameters:	inputs (dict) – Dictionary of the inputs to the tensorflow keras model
Returns:	train_features (tf.Tensor) – Dense tensor object that is used for training metadata_features (dict) – Dictionary of feature tensors that can be used for computing custom metrics and losses

UnivariateInteractionModel¶

class ml4ir.base.model.scoring.interaction_model.UnivariateInteractionModel(feature_config: ml4ir.base.features.feature_config.FeatureConfig, tfrecord_type: str, feature_layer_keys_to_fns: dict = {}, max_sequence_size: int = 0, file_io: ml4ir.base.io.file_io.FileIO = None)¶

Bases: ml4ir.base.model.scoring.interaction_model.InteractionModel

Defines an interaction model that configures feature layer operations on individual features

Constructor for instantiating a UnivariateInteractionModel

Parameters:

feature_config (FeatureConfig object) – FeatureConfig object that defines list of model features and the feature transformation functions to be used on each
tfrecord_type ({"example", "sequence_example"}) – Type of TFRecord protobuf being used for model training
feature_layer_keys_to_fns (dict) – Dictionary of custom feature transformation functions to be applied on the input features
max_sequence_size (int, optional) – Maximum size of the sequence in SequenceExample protobuf
file_io (FileIO object) – FileIO object that handles read write operations

feature_layer_op(inputs: Dict[str, tensorflow.python.keras.engine.input_layer.Input])¶

Convert input tensorflow features into numeric train features and metadata features by applying respective feature transformation functions as specified in the FeatureConfig

Parameters:	inputs (dict) – Dictionary of the inputs to the tensorflow keras model
Returns:	train_features (dict) – Dict of feature tensors that are used for training metadata_features (dict) – Dictionary of feature tensors that can be used for computing custom metrics and losses

transform_features_op(train_features: Dict[str, tensorflow.python.framework.ops.Tensor], metadata_features: Dict[str, tensorflow.python.framework.ops.Tensor])¶

Convert train and metadata features which have feature layer functions applied to them into dense numeric tensors. Sorts the features by name and concats the individual features into a dense tensor.

Parameters:	inputs (dict) – Dictionary of the inputs to the tensorflow keras model
Returns:	train_features (tf.Tensor) – Dense tensor object that is used for training metadata_features (dict) – Dictionary of feature tensors that can be used for computing custom metrics and losses

feature_layer¶

class ml4ir.base.features.feature_layer.FeatureLayerMap¶

Bases: object

Class defining mapping from keys to feature layer functions

Define ml4ir’s predefined feature transformation functions

add_fn(key, fn)¶

Add custom new function to the FeatureLayerMap

Parameters:	key (str) – name of the feature transformation function fn (tf.function) – tensorflow function that transforms input features

add_fns(keys_to_fns_dict)¶

Add custom new functions to the FeatureLayerMap

Parameters:	keykeys_to_fns_dict (dict) – Dictionary with name and definition of custom tensorflow functions that transform input features

get_fns()¶

Get all feature transformation functions

Returns:	Dictionary of feature transformation functions
Return type:	dict

get_fn(key)¶

Get feature transformation function using the key

Parameters:	key (str) – Name of the feature transformation function to be fetched
Returns:	Feature transformation function
Return type:	tf.function

pop_fn(key)¶

Get feature transformation function using the key and remove from FeatureLayerMap

Parameters:	key (str) – Name of the feature transformation function to be fetched
Returns:	Feature transformation function
Return type:	tf.function

ml4ir.base.features.feature_layer.define_feature_layer(feature_config: ml4ir.base.features.feature_config.FeatureConfig, tfrecord_type: str, feature_layer_map: ml4ir.base.features.feature_layer.FeatureLayerMap, file_io: ml4ir.base.io.file_io.FileIO)¶

Defines a feature layer function that works on keras.Inputs

Parameters:

feature_config (FeatureConfig object) – FeatureConfig object that defines the feature transformation function to be applied to each feature
tfrecord_type ({"example", "sequence_example"}) – String defining the TFRecord type of the data
feature_layer_map (FeatureLayerMap) – FeatureLayerMap object mapping custom function names to function definition

Returns:

tensorflow op that is the feature layer by collecting all the
feature transformation functions assigned to the respective
keras.Inputs features as specified by the FeatureConfig

Notes

Use feature_layer_map to define custom functions when using ml4ir as a library