TensorFlow Sigmoid activation function as output layer - value interpretation

Question

My TensorFlow model has the following structure. It aims to solve a binary classification problem where the labels are either 0 or 1. The output layer uses a sigmoid activation function with 1 output.

model = keras.Sequential([
    layers.Dense(10, activation='relu', input_shape=[len(train_dataset.keys())]),
    layers.Dense(1, activation='sigmoid')
  ])
  optimizer = 'adam'

  model.compile(loss='binary_crossentropy',
                optimizer=optimizer,
                metrics=[tf.keras.metrics.Precision(), tf.keras.metrics.Recall(), tf.keras.metrics.Accuracy()])

The output given is an array of dtype=float32 numbers that lie between 0 and 1.

array([[9.5879245e-01],
       [3.6847022e-01],
       [3.4174323e-04],
       ...,
       [2.6283860e-03],
       [3.2045375e-04],
       [1.0798702e-03]], dtype=float32)

The Tensorflow tutorials state that "Using the sigmoid activation function, this value is a float between 0 and 1 and represents a probability, or confidence level". https://www.tensorflow.org/tutorials/keras/text_classification

My question is:

Do I interpret the float values from output as:

1. "How likely it belongs to the first class label - in this case class 0 is my first class label?" e.g. model.predict() yields 0.99998 and therefore has a 99% chance of belonging to my first class label (class 0) and 1% belonging to the other class (class 1).

or

2. "The closer the output from model.predict() is to 0.0 the more likely it is class 0 and the closer the output is to 1.0 the more likely it is class 1"

Answer 1

In short, value of model.predict() function is interpreted as mentioned in option 2.

In order to clarify, let's assume we are talking about spam detection application. Label 0 represents that text/email is not spam and label 1 represents that text/email is spam.

Suppose, after running the function model.predict(), we get value 0.9899. Then we can interpret it as:

• chance of given text/email being spam is 98.99%
• chance of given text/email not being spam is 1.01%

Usually, when we talk about a binary classification problem, we will often use labels 0 & 1 to represent the classes. So the predicted value being closer to 0 means it is more likely to fall under the class 0 (in our example class 0 represents text/email not being spam.). Likewise, if the predicted value being closer to 1 means it is more likely to fall under the class 1 (in our example class 1 represents text/email being spam.)

Answer 2

please check this answer and see if it helps . Somehow I feel that you are getting the predict probability rather than the prediction itself

https://stackoverflow.com/questions/40002084/how-to-output-a-prediction-in-tensorflow