Deep learning with python code code#

If you havenвЂ™t, youвЂ™ll have a much easier time getting comfortable with those libraries in the future.ДЄnd throughout the book, Trask reminds you that practice makes perfect he encourages you to code your own neural networks by heart without copy-pasting anything. If youвЂ™ve already worked with other Python libraries such as Keras and PyTorch, youвЂ™ll find the final architecture to be quite familiar. YouвЂ™ll also get to craft your own version of convolutional neural networks (CNN) and recurrent neural networks (RNN).Д«y the end of the book, youвЂ™ll pack everything into a complete Python deep learning library, creating your own class hierarchy of layers, activation functions, and neural network architectures (youвЂ™ll need object-oriented programming skills for this part). The book features some popular regularization techniques such as early stopping and dropout. YouвЂ™ll also learn to compute gradients and propagate errors through layers to spread corrections across different neurons.ДЄs you get more comfortable with the basics of deep learning, youвЂ™ll get to learn and implement more advanced concepts. Again, youвЂ™ll implement everything yourself with the help of Numpy functions. With the basic building blocks of artificial neurons under your belt, youвЂ™ll start creating deep neural networks, which is basically what you get when you stack several layers of artificial neurons on top of each other.ДЄs you create deep neural networks, youвЂ™ll learn about activation functions and apply them to break the linearity of the stacked layers and create classification outputs. As you move from scalar to vector computations, youвЂ™ll shift from vanilla Python operations to Numpy, a library that is especially good at parallel computing and is very popular among the machine learning and deep learning community. YouвЂ™ll gradually add more complexity to your models, using multiple input dimensions, predicting multiple outputs, applying batch learning, adjusting learning rates, and more.ДЄnd youвЂ™ll implement every new concept by gradually adding and changing bits of Python code youвЂ™ve written in previous chapters, gradually creating a roster of functions for making predictions, calculating errors, applying corrections, and more. With a single neuron, single input, and single output, understanding and implementing the concept becomes very easy. YouвЂ™ll then measure the prediction error and apply gradient descent to tune the neuronвЂ™s weight in the right direction. In Grokking Deep Learning, your first artificial neuron will take a single input, multiply it by a random weight, and make a prediction. But writing everything in vanilla Python is excellent for learning the ins and outs of deep learning. This is not the most efficient way to do deep learning, because Python has many libraries that take advantage of your computerвЂ™s graphics card and parallel processing power of your CPU to speed up computations. You then implement the artificial neuron in plain Python code, without using any special libraries. Trask takes you through the basics of linear transformations, the main computation done by an artificial neuron. You start with developing a single artificial neuron, the most basic element of deep learning. In contrast, Grokking Deep Learning teaches you deep learning by building everything from scratch, line by line. Most deep learning books are based on one of several popular Python libraries such as TensorFlow, PyTorch, or Keras. The book wonвЂ™t make you a deep learning wizard (and it doesnвЂ™t make such claims), but it will set you on a path that will make it much easier to learn from more advanced books and courses. TraskвЂ™s book teaches you the fundamentals of deep learning without any prerequisites aside from basic math and programming skills. Most books on deep learning require a basic knowledge of machine learning concepts and algorithms. If I had to learn deep learning with Python all over again, I would start with Grokking Deep Learning, written by Andrew Trask. Knowing where to start, however, can help a lot in softening the learning curve. Even then, it can take more than a year of study and practice before you reach the point where you can start applying deep learning to real-world problems and possibly land a job as a deep learning engineer. You need a strong background in linear algebra and calculus, good Python programming skills, and a solid grasp of data science, machine learning, and data engineering. Teaching yourself deep learning is a long and arduous process.

Deep learning with python code series#

This article is part of вЂњAI educationвЂќ, a series of posts that review and explore educational content on data science and machine learning.