Curriculum — what I studied to construct the talents vital for passing the examination
It ought to be famous that earlier than I began learning for the examination, I had some hands-on expertise constructing a number of tasks with TensorFlow.
The skilled TensorFlow and deep studying practitioner will probably discover they will undergo the next curriculum at about the identical tempo I did (three weeks whole), possibly sooner.
The newbie will need to take as a lot time as wanted. Keep in mind: constructing any worthwhile ability takes time.
I’ve listed timelines, prices ($USD), and helpfulness degree (in direction of passing the examination) for every useful resource. The timelines are primarily based on my expertise.
If you wish to create a curriculum for your self, I’d advocate one thing like the next.
Notice: For paid assets, affiliate hyperlinks have been used. This doesn’t change the worth of the useful resource however should you do occur to buy one, I’ll obtain a portion of the fee: cash I exploit in direction of creating assets like this.
Helpfulness degree: Required.
This ought to be your first cease. It outlines the subjects which might be lined within the examination. Learn it after which learn it once more.
If you happen to’re new to TensorFlow and machine studying, you’ll probably learn this and get scared in any respect the completely different subjects. Don’t fear. The assets beneath will enable you to turn out to be accustomed to them.
Time: Three weeks (superior person) to three months (newbie).
Value: $59 monthly after a 7-day free trial, monetary assist obtainable by means of utility. If you happen to can’t entry Coursera, see the equal free model on YouTube.
Helpfulness degree: 10/10.
That is essentially the most related useful resource to the examination (and getting began with TensorFlow normally). The cautious pupil will discover the TensorFlow Certification handbook and the define of this specialization are virtually an identical.
It’s taught by Laurence Moroney and Andrew Ng, two titans of TensorFlow and machine studying and if I needed to solely select one useful resource to arrange for the examination, this is able to be it.
I appreciated the quick video format and concentrate on hands-on examples as quickly as potential. The a number of code notebooks on the finish of every part have been must-haves for any sensible learner.
A tip for the programming workouts: don’t simply fill within the code gaps, write your entire factor out your self.
Time: Three weeks (studying cowl to cowl, no workouts) — Three months (studying cowl to cowl and doing the workouts).
Value: Value varies on Amazon however I picked up a tough copy for $55. You may see all of the code free of charge on GitHub.
Helpfulness degree: 7/10 (solely as a result of some chapters aren’t related to the examination).
At 700+ pages, this e-book covers mainly all of machine studying and thus, some subjects which aren’t related to the examination. Nevertheless it’s a must-read for anybody all in favour of setting themselves a stable basis for a future in machine studying and never simply to go an examination.
If you happen to’re new to machine studying, you’ll in all probability discover this e-book laborious to learn (to start with). Once more, to not fear, you’re not in a rush, studying helpful abilities takes time.
Put it this fashion, if you would like an concept of the standard of the e-book, I learn the primary version throughout morning commutes to my machine studying engineer job. And I can inform you, as a rule, I’d find yourself utilizing precisely what I learn within the e-book throughout the day.
The 2nd version is not any completely different, besides it’s been up to date to cowl the newest instruments and strategies, specifically TensorFlow 2.x — what the examination relies on.
If you happen to’re solely after related chapters to the examination, you’ll need to learn:
- Chapter 10: Introduction to Synthetic Neural Networks with Keras
- Chapter 11: Coaching Deep Neural Networks
- Chapter 12: Customized Fashions and Coaching with TensorFlow
- Chapter 13: Loading and Preprocessing Information with TensorFlow
- Chapter 14: Deep Pc Imaginative and prescient Utilizing Convolutional Neural Networks
- Chapter 15: Processing Sequences Utilizing RNNs and CNNs
- Chapter 16: Pure Language Processing with RNNs and Consideration
However for the intense pupil, I’d counsel the entire e-book and the workouts (possibly not all, however decide and the select those which swimsuit spark your pursuits most).
Time: 3-hours (I solely watched Three lectures) — 24-hours (1-hour per lecture, plus 1-hour overview every).
Helpfulness degree: 8/10.
World-class deep studying info from a world-class college, oh and did I point out? It’s free.
The primary Three lectures, deep studying (normally), Convolutional Neural Networks (often used for laptop imaginative and prescient), and Recurrent Neural Networks (often used for textual content processing) are essentially the most related to the examination.
However once more, for the keen learner, going by means of the entire course wouldn’t be a nasty concept.
You’ll want to take a look at the labs and code they provide on GitHub, particularly the Introduction to TensorFlow one. And once more, I can’t stress the significance of writing the code your self.
Time: 3-hours (relying on how briskly your laptop is).
Helpfulness degree: 10/10 (utilizing PyCharm is a requirement).
The examination takes place in PyCharm (a Python improvement software). Earlier than the examination, I’d by no means used PyCharm. And it’s instructed you get at the very least considerably accustomed to it earlier than you begin.
So to familiarize myself with PyCharm, I went by means of their getting began collection on YouTube which was very easy, “right here’s what this button does.”
However the principle assessments have been ensuring TensorFlow 2.x ran with none points and my laptop may run deep neural networks in a good time (my MacBook Professional doesn’t have a Nvidia GPU).
To check this, I replicated the next two TensorFlow tutorials on my native machine:
Each of those labored superb regionally, nevertheless, as we’ll see beneath, as quickly as I began the examination, I bumped into a problem.
- deeplearning.ai movies on Coursera/YouTube — The examination is solely code-based (Python code) however if you wish to know what’s happening behind the scenes of the code you’re writing (linear algebra, calculus), I’d bounce out and in of movies right here as you see match. For instance, should you’re undecided what mini-batch gradient descent is, search “deeplearning.ai mini-batch gradient descent.”
- TensorFlow documentation — If you happen to’re going to be a TensorFlow practitioner, you’re going to wish to have the ability to learn the documentation. If you happen to don’t perceive one thing, write the code and remark it your self.
- Coding with TensorFlow on YouTube (playlist) — Many of the TensorFlow in Observe Coursera Specialization in video collection on YouTube, taught by the identical teacher as nicely.