TensorFlow is an open source software library developed by Google for numerical computation using data flow graphs. The R interface for TensorFlow is developed by RStudio, which provides an interface for three TensorFlow APIs:
- Keras
- Estimator
- Core
The keras, tfestimators, and tensorflow packages provide R interfaces to the aforementioned APIs, respectively. Keras and Estimator are high-level APIs, while Core is a low-level API that offers full access to the core of TensorFlow. In this recipe, we will demonstrate how we can build and train deep learning models using Keras.
Keras is a high-level neural network API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. The R interface for Keras uses TensorFlow as its default backend engine. The keras package provides an R interface for the TensorFlow Keras API. It lets you build deep learning models in two ways, sequential and functional, both of which will be described in the following sections.
