![]() ![]() We’ll use the sequential API for our models. The latter is useful in more advanced scenarios such as networks featuring non-sequential topologies or shared layers. The former is simpler and is sufficient for most neural networks. Keras offers two APIs: a sequential API and a functional API. Even Google recommends using the Keras API. (Keras can also use CNTK and Theano as back ends, but development has been halted on those frameworks and they are rarely used on new projects.) Keras began life as a separate project in 2015 but was merged into TensorFlow in 2019. Any Keras code that you write ultimately executes in TensorFlow. Another library named Keras provides a simplified Python interface to TensorFlow and has emerged as the Scikit of deep learning. The learning curve for TensorFlow is rather steep. TensorFlow lets you define directed graphs that in turn define how tensors are computed. The 3-layer perceptron featured in my previous post takes a 1D tensor containing two values as input, transforms it into a 1D tensor containing three values, and produces a 0D tensor as output. A neural network is basically a workflow for transforming tensors. Tensors can represent scalar values (0-dimensional tensors), vectors (1D tensors), matrices (2D tensors), and so on. It is a framework for performing fast mathematical operations at scale using tensors, which are simply arrays. TensorFlow isn’t limited to building neural networks. But the library that most of the world has settled on for building neural networks is TensorFlow, an open-source framework created by Google that was released under the Apache License 2.0 in 2015. Deep learning isn’t hard, either, thanks to libraries such as the Microsoft Cognitive Toolkit (CNTK), Theano, and PyTorch. The reason it’s not hard is libraries such as Scikit-learn and ML.NET, which reduce complex mathematical manipulations to a few lines of code. Machine learning isn’t hard when you have a properly engineered dataset to work with.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |