"""Module containing an interface to trained Keras Tensorflow model."""
import tensorflow as tf
from tensorflow import keras
from dice_ml.model_interfaces.base_model import BaseModel
[docs]
class KerasTensorFlowModel(BaseModel):
def __init__(self, model=None, model_path='', backend='TF1', func=None, kw_args=None):
"""Init method
:param model: trained Keras Sequential Model.
:param model_path: path to trained model.
:param backend: "TF1" for TensorFlow 1 and "TF2" for TensorFlow 2.
:param func: function transformation required for ML model. If func is None, then func will be the identity function.
:param kw_args: Dictionary of additional keyword arguments to pass to func. DiCE's data_interface is appended to the
dictionary of kw_args, by default.
"""
super().__init__(model, model_path, backend, func, kw_args)
[docs]
def load_model(self):
if self.model_path != '':
self.model = keras.models.load_model(self.model_path)
[docs]
def get_output(self, input_tensor, training=False, transform_data=False):
"""returns prediction probabilities
:param input_tensor: test input.
:param training: to determine training mode in TF2.
:param transform_data: boolean to indicate if data transformation is required.
"""
if transform_data or not tf.is_tensor(input_tensor):
input_tensor = tf.constant(self.transformer.transform(input_tensor).to_numpy(), dtype=tf.float32)
if self.backend == 'TF2':
return self.model(input_tensor, training=training)
else:
return self.model(input_tensor)
[docs]
def get_gradient(self, input_instance):
# Future Support
raise NotImplementedError("Future Support")
[docs]
def get_num_output_nodes(self, inp_size):
temp_input = tf.convert_to_tensor([tf.random.uniform([inp_size])], dtype=tf.float32)
return self.get_output(temp_input)