In this section, we will discuss gradient descent-based optimization options that are provided by TensorFlow. Initially, it will not be clear which optimization option you should use, but as and when you know the actual logic of the DL algorithm, it will became much clearer to you.
We use a gradient descent-based approach to develop an intelligent system. Using this algorithm, the machine can learn how to identify patterns from the data. Here, our end goal is to obtain the local minimum and the objective function is the final prediction that the machine will make or result that is generated by the machine. In the gradient descent-based algorithm, we are not concentrating on how to achieve the best final goal for our objective function in the first step, but we will iteratively or repeatedly take small steps and select the intermediate best option...