Lesson 1: Practical Deep Learning for Coders



NB: All files referenced in the video have actually moved from platform.ai to files.fast.ai. We recommend going to http://course.fast.ai/lessons/lesson1.html to enjoy this video, in order to access the composed notes, tasks, assistance online forums, and so on. If you have any concerns or require assistance at any point, please ask at http://forums.fast.ai, where countless trainees are assisting each other with this course.
Invite to the very first complete lesson of Practical Deep Learning For Coders! Prior to you begin this lesson, make sure to have actually finished setup of your deep knowing server. See the AWS Lesson to find out ways to do this, if you have not currently.

Each lesson page consists of connect to course notes, online forum conversation, and (most notably) a wiki page. Almost all the individuals in the initial in-person course stated that they discovered these resources essential for effectively finishing the course. Be sure to make the many of them! And make sure to thoroughly check out the Getting Started page to learn how this course is developed and ways to get the most out of it. (Also, apologies that the concerns from the audience are difficult to hear – we get an unique audience mic from lesson 3 onwards which solves that issue.).

SUMMARY.

The 30 minute summary video presents you to the course and discusses ways to get the most from each lesson. We likewise hand down some ideas from previous trainees.

The lesson video begins with an extremely quick summary of exactly what deep finding out is, and why it matters, then talks about ways to access the apply for this lesson. Keep in mind that for this MOOC you will most likely discover it simpler to utilize git rather. The Getting Started page discusses how. We reveal how to begin, stop, and handle your AWS circumstances, how to copy information to it, and so forth (if you’re currently familiar with AWS you can most likely go through this part relatively rapidly).

The next point talked about is ways to the information for this lesson (and undoubtedly all the computer system vision jobs we’ll deal with) has to be structured. If your information is not structured properly you will not be able to train any designs, this is the most crucial action for you to finish–.

We get into our very first deep knowing design (http://wiki.fast.ai has a fully-hyperlinked timeline for each lesson, so you can quickly leap to any area of any video). We find out ways to categorize canines from felines. Instead of comprehending the mathematical information of how this works, we begin by finding out the nuts and bolts of ways to get the computer system to finish the job, utilizing ‘tweak’, maybe the most crucial ability for any deep knowing professional. When we’ve developed a design that can categorize canines from felines, we then take an action back and learn more about the initial “pre-trained” design that we began with, and analyze exactly what it can do with no fine-tuning. Next lesson, we’ll learn more about how fine-tuning really works “behind the scenes”.