{"id":22636,"date":"2021-02-19T09:39:00","date_gmt":"2021-02-19T09:39:00","guid":{"rendered":"https:\/\/www.experfy.com\/blog\/if-know-nothing-deep-learning-with-python-start-here\/"},"modified":"2023-09-04T18:19:54","modified_gmt":"2023-09-04T18:19:54","slug":"if-know-nothing-deep-learning-with-python-start-here","status":"publish","type":"post","link":"https:\/\/www.experfy.com\/blog\/ai-ml\/if-know-nothing-deep-learning-with-python-start-here\/","title":{"rendered":"If You Know Nothing About Deep Learning With Python, Start Here"},"content":{"rendered":"\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Teaching yourself deep learning is a long and arduous process. You need a strong background in\u00a0linear algebra and calculus<\/a>, good Python programming skills, and a solid grasp of data science, machine learning, and data engineering. Even then, it can take more than a year of study and practice before you reach the point where you can start applying\u00a0deep learning<\/a>\u00a0to real-world problems and possibly land a job as a deep learning engineer.<\/p>\n

Knowing where to start, however, can help a lot in softening the learning curve. If I had to learn deep learning with Python all over again, I would start with\u00a0Grokking Deep Learning<\/em><\/a>, written by Andrew Trask. Most books on deep learning require a basic knowledge of\u00a0machine learning<\/a>\u00a0concepts and algorithms. Trask\u2019s book teaches you the fundamentals of deep learning without any prerequisites aside from basic math and programming skills.<\/p>\n

The book won\u2019t make you a deep learning wizard (and it doesn\u2019t 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.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Building an artificial neuron in Python<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"If\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Most deep learning books are based on one of several popular Python libraries such as TensorFlow, PyTorch, or Keras. In contrast, Grokking Deep Learning<\/em> teaches you deep learning by building everything from scratch, line by line.<\/p>\n

You start with developing a single artificial neuron,\u00a0the most basic element of deep learning<\/a>. Trask takes you through the basics of linear transformations, the main computation done by an artificial neuron. <\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

You then implement the artificial neuron in plain Python code, without using any special libraries.<\/p>\n

This is not the most efficient way to do deep learning, because Python has many libraries that take advantage of your computer\u2019s graphics card and parallel processing power of your CPU to speed up computations. But writing everything in vanilla Python is excellent for learning the ins and outs of deep learning.<\/p>\n

In Grokking Deep Learning<\/em>, your first artificial neuron will take a single input, multiply it by a random weight, and make a prediction. You\u2019ll then measure the prediction error and apply gradient descent to tune the neuron\u2019s weight in the right direction. With a single neuron, single input, and single output, understanding and implementing the concept becomes very easy. You\u2019ll gradually add more complexity to your models, using multiple input dimensions, predicting multiple outputs, applying batch learning, adjusting learning rates, and more.<\/p>\n

And you\u2019ll implement every new concept by gradually adding and changing bits of Python code you\u2019ve written in previous chapters, gradually creating a roster of functions for making predictions, calculating errors, applying corrections, and more. As you move from scalar to vector computations, you\u2019ll 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.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Deep neural networks with Python<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\"deep\t\t\t\t\t\t\t\t\t\t\t
Deep neural networks use multiple layers of parameters to map input data to outputs<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

With the basic building blocks of artificial neurons under your belt, you\u2019ll start creating\u00a0deep neural networks<\/a>, which is basically what you get when you stack several layers of artificial neurons on top of each other.<\/p>\n

As you create deep neural networks, you\u2019ll learn about activation functions and apply them to break the linearity of the stacked layers and create classification outputs. Again, you\u2019ll implement everything yourself with the help of Numpy functions. You\u2019ll also learn to compute gradients and propagate errors through layers to spread corrections across different neurons.<\/p>\n

As you get more comfortable with the basics of deep learning, you\u2019ll get to learn and implement more advanced concepts. The book features some popular regularization techniques such as early stopping and dropout. You\u2019ll also get to craft your own version of\u00a0convolutional neural networks<\/a>\u00a0(CNN) and\u00a0recurrent neural networks<\/a>\u00a0(RNN).<\/p>\n

By the end of the book, you\u2019ll pack everything into a complete Python deep learning library, creating your own class hierarchy of layers, activation functions, and neural network architectures (you\u2019ll need object-oriented programming skills for this part). If you\u2019ve already worked with other Python libraries such as Keras and PyTorch, you\u2019ll find the final architecture to be quite familiar. If you haven\u2019t, you\u2019ll have a much easier time getting comfortable with those libraries in the future.<\/p>\n

And 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.  <\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Code library is a bit cumbersome<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Not everything about\u00a0Grokking Deep Learning<\/em>\u00a0is perfect. In a\u00a0previous post<\/a>, I said that one of the main things that define a good book is the code repository. And in this area, Trask could have done a much better job.<\/p>\n

The\u00a0GitHub repository<\/a>\u00a0of\u00a0Grokking Deep Learning\u00a0<\/em>is rich with Jupyter Notebook files for every chapter. Jupyter Notebook is an excellent tool for learning Python machine learning and deep learning. However, the strength of Jupyter is in breaking down code into several small cells that you can execute and test independently. Some of\u00a0Grokking Deep Learning<\/em>\u2019s notebooks are composed of very large cells with big chunks of uncommented code.<\/p>\n

This becomes especially problematic in the later chapters, where the code becomes longer and more complex, and finding your way in the notebooks becomes very tedious. As a matter of principle, the code for educational material should be broken down into small cells and contain comments in key areas.<\/p>\n

Also, Trask has written the code in Python 2.7. While he has made sure that the code also works smoothly in Python 3, it contains old coding techniques that have become deprecated among Python developers (such as using the \u201cfor i in range(len(array))<\/em>\u201d paradigm to iterate over an array).<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

The broader picture of artificial intelligence<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"If\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Trask has done a great job of putting together a book that can serve both newbies and experienced Python deep learning developers who want to fill the gaps in their knowledge.<\/p>\n

But as Tywin Lannister says (and every engineer will agree), \u201cThere\u2019s a tool for every task, and a task for every tool.\u201d Deep learning isn\u2019t a magic wand that can solve every AI problem. In fact, for many problems, simpler machine learning algorithms such as linear regression and decision trees will perform as well as deep learning, while for others,\u00a0rule-based techniques<\/a>\u00a0such as regular expressions and a couple of if-else clauses will outperform both.<\/p>\n

The point is, you\u2019ll need a full arsenal of tools and techniques to solve AI problems. Hopefully, Grokking Deep Learning<\/em> will help get you started on the path to acquiring those tools.<\/p>\n

You can also pick up a lot of knowledge browsing machine learning and deep learning forums such as the\u00a0r\/MachineLearning<\/a>\u00a0and\u00a0r\/deeplearning<\/a>\u00a0subreddits, the\u00a0AI and deep learning Facebook group<\/a>, or by following AI researchers on Twitter.<\/p>\n

The AI universe is vast and quickly expanding, and there is a lot to learn. If this is your first book on deep learning, then this is the beginning of an amazing journey.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"

Teaching yourself deep learning is a long and arduous process. 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.<\/p>\n","protected":false},"author":109,"featured_media":18764,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[183],"tags":[1352,206,1353,408,114],"ppma_author":[1946],"class_list":["post-22636","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-ml","tag-calculus","tag-deep-learning","tag-linear-algebra","tag-programming","tag-python"],"authors":[{"term_id":1946,"user_id":109,"is_guest":0,"slug":"ben-dickson","display_name":"Ben Dickson","avatar_url":"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/04\/medium_8aaf6bea-c4c1-455f-8156-8007d70910f8-150x150.jpg","user_url":"https:\/\/bdtechtalks.com\/","last_name":"Dickson","first_name":"Ben","job_title":"","description":"Ben Dickson is an experienced software engineer and tech blogger. He contributes regularly to major tech websites such as the Next Web, the Daily Dot, PCMag.com, Cointelegraph, VentureBeat, International Business Times UK, and The Huffington Post."}],"_links":{"self":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/22636","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/users\/109"}],"replies":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/comments?post=22636"}],"version-history":[{"count":4,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/22636\/revisions"}],"predecessor-version":[{"id":32159,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/22636\/revisions\/32159"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media\/18764"}],"wp:attachment":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media?parent=22636"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/categories?post=22636"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/tags?post=22636"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/ppma_author?post=22636"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}