Lesson 1 Overview

00:00:00.000 | Hi, and welcome to lesson one of this deep learning MOOC. Thanks for joining us. I'm Jeremy Howard and I'm Rachel Thomas

00:00:08.000 | We're the people who put this together. You'll be seeing my face in front of the camera most of the time

00:00:14.640 | You'll be hearing Rachel's voice however, and I'll be asking questions that were coming through a slack channel asked by students in the in-person version of this course

00:00:23.040 | So we thought before we started we should tell you a little bit about what to expect and maybe you get to know us a little bit

00:00:29.760 | My background is really in

00:00:31.760 | Coding and data I spent 10 years of management consulting and 10 years running startups throughout that time

00:00:40.320 | I've been using data and machine learning to try to solve problems

00:00:43.660 | My background is a lot more academic and theoretical

00:00:47.240 | I have a PhD in math and then I worked as a quant later as a software engineer and data scientist at Uber

00:00:54.960 | One of the most fun and exciting parts of my life was when I spent some time really competing heavily in Kaggle competitions

00:01:01.920 | I was really pleased to win some of those competitions and get to the top of the leaderboard

00:01:06.520 | And I'm hoping to show you guys during this course some of the techniques that I used to do that

00:01:11.720 | I think the techniques that allow you to win Kaggle competitions are the same as the techniques that allow you to great

00:01:17.180 | Great results on your own models in solving your own problems

00:01:23.680 | we also both love teaching and so I taught calculus one and two when I was in graduate school and

00:01:30.160 | Then I later left my job as a software engineer to teach full-stack software development women at Hackbright Academy for a year and a half

00:01:37.200 | I think that's really cool Rachel was a quant and then she works as both a data scientist and a full-stack engineer at Uber

00:01:43.960 | But she realized that one of the highest leverage things you can do is to teach and it's great fun, too

00:01:50.740 | I feel the same way even when I was running startups. I was creating

00:01:55.280 | Course content online for example on the left here is a angular JS tutorial that I originally created for my colleagues at Kaggle

00:02:04.400 | But I recorded it and put it online and it's had over 200,000 views

00:02:08.300 | Makes me feel really good to know that people are learning

00:02:11.980 | From some of the things that I found really helpful myself

00:02:19.080 | So this is a quote from Paul Lockhart who was a

00:02:22.180 | He was actually working as a primary school math teacher

00:02:26.000 | Got his PhD in math at Columbia and became a math professor at Brown and then left to go back to teaching primary school

00:02:32.100 | And he's written a wonderful essay called a mathematician's lament on everything. That's horrible about how mathematics is taught in the United States

00:02:39.300 | Yeah, I think that that essay has been really influential to both Rachel and I

00:02:44.520 | Although Rachel stuck with her math education for decades longer than I did

00:02:49.040 | We both definitely felt like modern mathematics education is not done. Well

00:02:54.280 | Paul Lockhart uses a wonderful analogy about imagine if with music

00:02:59.560 | We didn't allow children to sing or play instruments until they had spent

00:03:04.880 | Years and even decades studying set theory and music notation and could transcribe scales and only then once they were in their 20s

00:03:13.480 | We would let them create music

00:03:15.480 | He says that's exactly what we're doing with mathematics, but that we should let people kind of

00:03:19.880 | Play and create and build patterns with it and something very similar happens with deep learning and how it's taught

00:03:26.480 | In fact, one of my heroes is a guy called David Perkins who at Harvard has created some

00:03:33.540 | Really interesting research about effective educational techniques and he has a very similar analogy to Paul Lockhart

00:03:40.440 | But he talks about baseball imagine if the way you learn baseball was that you never saw a game of baseball

00:03:47.360 | But instead you learned about how to stitch a baseball and you like the physics of a parabola

00:03:51.920 | And you learn every aspect of baseball and then after 20 years of study could be considered good enough to go and actually watch

00:03:58.640 | your first game

00:04:00.040 | We tend to think that this is rather the way that most mathematics perhaps particularly including deep learning is really taught

00:04:06.600 | so we

00:04:10.000 | Decided when we set up our research lab fast AI that the first thing we would do would be to try and

00:04:15.000 | Fill this need and particularly we decided to focus on deep learning because we both think that deep learning is

00:04:21.920 | The most exciting technology that we have ever seen

00:04:26.160 | We think it's going to be more transformative than even than the internet and so the more people who can participate the better

00:04:32.880 | Andrew Ng has called it the new electricity but kind of to say that it's going to have the impact on society that electricity has

00:04:40.040 | so

00:04:41.880 | Some other kinds of problems we've seen with technical teaching for these and I just want to say we're introducing this to tell you

00:04:48.200 | That this course is taught in a very different style

00:04:50.160 | And so we want to kind of set your expectations ahead of time and motivate why it's so different how we teach it here

00:04:55.760 | And one is that a lot of them existing deep learning materials are very math centric and even as a mathematician and someone who loves math

00:05:04.360 | I found them to be pretty unhelpful for actually building and creating practical applications

00:05:11.040 | In fact every time I see somebody ask on a forum or on Hacker News or whatever

00:05:15.580 | What do I need in order to get into deep learning a whole bunch of people reply by saying well first?

00:05:21.280 | You need five years of real analysis and vector analysis

00:05:25.000 | And then you need to study probability and statistics and blah blah blah blah blah and it really comes across to me as

00:05:31.800 | Something which is all about being exclusive rather than inclusive. So that's why we have this little

00:05:37.880 | Thing making your own at some call again is kind of our

00:05:42.760 | slogan we're all about not being exclusive but about making things as simple as possible, but never about

00:05:52.640 | Dumbing it down, right?

00:05:55.280 | another way that

00:05:57.760 | Kind of technical education fails is what David Perkins and the Harvard professor Jeremy mentioned a moment ago

00:06:04.440 | Calls elementitis and that's that often math does this so much

00:06:09.880 | It teaches kind of each separate element and it's only at the end when you've learned all the elements

00:06:15.600 | Needed that you can put them together and see the whole thing and that's kind of what was going on in that baseball analogy

00:06:20.840 | And that happens in a lot of deep learning

00:06:22.440 | It's like, you know, we need to teach you probability theory and we need to teach you information theory and only way later on

00:06:28.200 | Are we going to let you put it together?

00:06:29.440 | You can think of it as being depth first rather than breadth first if you like

00:06:33.560 | So the traditional depth first approach means that you as a student have to trust that at some point all these things are going

00:06:39.760 | To come together and turn into something that's genuinely useful

00:06:43.480 | I think with this breadth first approach you still have to trust but it's different kind of trust

00:06:49.000 | Which is that it's okay that when we first show you an end-to-end process that you don't deeply understand every part

00:06:56.720 | But that you are able to actually do useful things from the very first lesson and that as the lessons go along

00:07:03.980 | You're going to get more and more in-depth understanding of each piece and two ways that the elementitis or the depth first

00:07:10.880 | Approach fails our one is motivation

00:07:13.760 | A lot of students kind of give up because they don't have the motivation of seeing how are these going to fit together?

00:07:18.040 | And then secondly, it's harder to get that like you don't have the context when you're learning all these discrete elements and you

00:07:24.360 | Can't learn how they're going to fit into the process until later

00:07:27.800 | Right and in fact this goes together with the idea of using a code centric

00:07:32.940 | Approach and sort of a math centric approach with a code centric approach and looking at the whole game

00:07:38.080 | That is an end-to-end machine learning process from the very start. It means that you can do experiments

00:07:43.680 | You can actually run experiments and see what goes in and out of each part of the system and build up that intuition

00:07:49.600 | And if this whole game analogy intrigues you David Perkins has a book called making learning hole where he goes into a lot more detail

00:07:57.040 | About it. Love that book

00:07:59.040 | So then not not only are we going to be showing you end-to-end processes from this very first lesson

00:08:04.960 | But these processes are going to not going to just end up with good enough results nearly all of the deep learning

00:08:11.680 | educational materials I've seen so far get you to a point where

00:08:15.560 | You can kind of get an okay-ish result now

00:08:19.060 | The whole point of deep learning is that you can get state-of-the-art results and so in the very first piece of code

00:08:24.680 | We're going to run we're going to run a piece of code which gives you a state-of-the-art result

00:08:28.560 | We know something as a state-of-the-art result if it is better than other approaches that people have tried

00:08:35.600 | The best way to know that is to try things on a Kaggle competition

00:08:39.920 | Having been the president and chief scientist of Kaggle

00:08:42.320 | I saw again and again that every Kaggle competition beat all previous academic state state-of-the-art results

00:08:48.300 | So very often in this course

00:08:50.800 | We're actually going to use Kaggle benchmarks and see if we can beat them because we know if we can then that's truly a world

00:08:57.240 | best

00:08:59.240 | So this is from actually a very good book

00:09:04.960 | It's the Ian Goodfellow Yoshua Bengio deep learning book

00:09:08.200 | But it's a very good math book which teaches you the math of deep learning and so in this book when they say

00:09:15.200 | Here is how we gain some intuition in how to back propagation through time works. This is how they develop intuition

00:09:22.380 | Rachel is a math PhD. Did you find this helped your intuitions?

00:09:26.240 | We'll have a very different approach to intuition

00:09:29.960 | So this is a good book if you're interested in math and theorems in this course

00:09:33.720 | We're really going to be focused on code

00:09:35.720 | In fact, this is what Rachel and I put together when we were trying to explain back prop and specifically

00:09:42.920 | Stochastic gradient descent and the use of back prop there was we created a spreadsheet

00:09:47.320 | And we found each time that we taught our students in the in-person course through a spreadsheet

00:09:53.840 | They could see every single piece of what was going on every single intermediate result

00:09:58.760 | And it was very easy for them to experiment with and so one of the unusual things we do is that you'll see that

00:10:04.760 | nearly every major

00:10:07.360 | Idea is presented at some point using a spreadsheet. We present it in many different ways, but spreadsheets

00:10:13.400 | Diagrams and code are three of the key ways that we present these ideas

00:10:18.080 | I believe this is the first deep learning course in the world to implement

00:10:22.440 | convolutional neural network in an Excel spreadsheet and also as you see from this page not just stochastic gradient descent

00:10:28.640 | But at a grad or a mess prop Adam and even Eve which just came out a few weeks ago

00:10:34.240 | or modern examples of

00:10:36.760 | Accelerated SGD approaches

00:10:40.520 | So I think everything you really need to know about the course comes in this very first piece of code that you see

00:10:48.560 | And this very first piece of code that you see you can see that there's a number of things going on

00:10:53.240 | The first is that this piece of code shows not just how to complete a project

00:10:59.720 | But how to get a state-of-the-art result on a project this particular piece of code gives you 97%

00:11:06.520 | accuracy in determining cats versus dogs

00:11:09.340 | As recently as about five years ago the state-of-the-art for this particular

00:11:14.880 | Problem was about 80% accuracy

00:11:18.520 | Um, it's also an example of showing why working with code is so interesting

00:11:25.180 | rather than showing math

00:11:28.000 | What we're showing here is some working code and I'll give an example of what that means you can do

00:11:33.800 | So the code environment that we're working in is something called Jupiter notebook

00:11:38.140 | And you'll be using this in every single lesson throughout the course and in Jupiter notebook as you can see we provide you with

00:11:47.600 | Pros and information about what's going on and we draw pictures and at any point in time

00:11:53.280 | you can take a look at one of these results and you can

00:11:57.580 | You can take a look at one of these results and you can look to see what's going on behind the scenes

00:12:05.360 | So for example in this case

00:12:06.760 | We're running something called VGG dot predict and we're getting back some probabilities and you might wonder well

00:12:13.100 | What's VGG dot predict actually doing so at any time you can take anything and put two question marks on the front and

00:12:19.480 | Run that piece of code and it will actually show you

00:12:23.280 | The full documentation and source code of what you just ran now in this case

00:12:28.880 | It's actually running a function that we wrote for you

00:12:31.920 | One of the other different things about this course is that we're not just showing you how existing libraries work

00:12:38.600 | But every time we found that using somebody else's library takes more than four or five lines of code

00:12:44.660 | We would make sure we found a way to do it easier

00:12:47.860 | So generally speaking we show you how to do things in one line of code and then you can look behind the scenes and see

00:12:54.040 | What the lines of code are actually doing?

00:12:56.040 | so for example in this case the predict method is running some other predict method called model dot predict and

00:13:04.280 | So then what I always encourage people to do is to do some experiments. So what does model dot predict actually do?

00:13:11.620 | one thing that you can do in Jupyter notebook at any time is to press shift tab a couple of times and

00:13:18.640 | When you press shift tab the first time it pops up is tells you what?

00:13:23.400 | Parameters you need to pass this method and it also tells you what the method actually does

00:13:32.080 | If you press it three times it then gives you additional information about what each of those arguments are and what they're expecting and what it

00:13:39.320 | returns

00:13:40.240 | So it's really nice that using this method you can find out exactly what's going on behind the scenes and do some experiments

00:13:47.840 | And so then for example, you could find out. Okay. Well, what is the shape? What is the size and shape of the array?

00:13:54.040 | That this thing returns

00:13:56.040 | What are the first four elements of the classes that are in this?

00:14:01.120 | object and so forth and this is really the way to

00:14:05.520 | Use this style of teaching effectively

00:14:09.400 | It's to have the code in front of you all of the time and in every line look and see what's being passed in

00:14:15.660 | What's coming out? What else could we do with that and then even look at the documentation?

00:14:22.000 | So VGG dot model is apparently a care us dot model dot sequential. So if we were to just copy that into

00:14:30.580 | Google

00:14:32.580 | Then we can click on the first item and find out exactly what's going on what is being used here

00:14:41.640 | What are the other methods that this could take and then we can try calling some of these other methods and see what kind of results

00:14:47.420 | We get so really what we're trying to do is in the two hours of each lesson

00:14:52.480 | We're trying to give you enough information to get you started with your own experiments

00:14:57.780 | We're not trying to teach you everything and we're certainly not assuming that the lesson can stand alone

00:15:03.760 | And we'll talk more about this in a moment

00:15:05.900 | But the videos are just a small part of the course but the IPY on notebooks and the code are a huge resource

00:15:11.140 | And we'll talk about some of the other resources that we have available for you

00:15:14.140 | But the important thing to realize with these

00:15:16.740 | Six lines of code is that you can run this for anything not just for dogs versus cats

00:15:21.880 | But these first six lines of code you learn

00:15:23.860 | We're actually as it says here work for any image recognition task with any number of categories

00:15:28.460 | So if you can get this far in today's lesson

00:15:31.420 | Then you've learned to do one of the most important types of computer vision

00:15:34.940 | Which is image classification or any number of categories for any type of images?

00:15:39.660 | As Rachel said we've actually run this course already

00:15:45.540 | Specifically what you're going to be seeing are the recorded lessons from an in-person course

00:15:50.900 | And we thought it'd be helpful for you to see what some of our students said about that in-person course because it might

00:15:56.740 | Help you to be a more effective learner

00:15:59.300 | And I do want to say I'm again because this course is taught in such a different way that

00:16:05.980 | It takes some faith kind of that this new technique is worth trying and kind of sticking with

00:16:12.220 | but you can see that

00:16:15.060 | almost all the students said that this was that the homework assignments were very helpful or

00:16:20.780 | Extremely helpful in understanding the material

00:16:24.420 | And the class resources which includes the wiki the scripts that we give you our forums our slack channel

00:16:31.460 | We're very helpful or extremely helpful and we want to mention and we wanted to mention that because Rachel and I are both being

00:16:38.140 | kind of Coursera addicts in the past and Udacity addicts and

00:16:42.180 | Generally speaking we all often watch a video at one and a half speed or two speed and just zip through them

00:16:47.920 | This is not designed to be possible to do that this way

00:16:52.240 | This is designed that you need to use the homework assignments and the class resources

00:16:57.220 | So as you can see from the people who have already been through this class

00:16:59.900 | They're actually finding that these are really important parts of the overall course

00:17:03.700 | As each video is giving you you're kind of seeing an end-to-end process of solving a real problem with deep learning

00:17:10.860 | And that means that there's not though a separate video on the kind of this is everything you need to know about

00:17:15.180 | AWS in your environment and this is everything you need to know about this piece of code

00:17:20.800 | But rather you're kind of seeing the end-to-end process, but you'll see it again and again throughout the lessons

00:17:26.620 | Now it's okay if you're coming into this course with either a very large amount or a very small amount of data science background

00:17:34.300 | Everybody in the in-person course simply had to have had at least a year of coding experience

00:17:40.500 | even with that very wide variety and background

00:17:43.220 | Nearly everybody said they found the pacing about right for them

00:17:47.900 | And the reason for that I think is that we really give people the ability to pick up as much as little as they want

00:17:54.140 | Through the forums if you want to dig very very deep into advanced topics you can or if absolutely everything is new to you

00:18:03.040 | Then that's fine, too

00:18:04.820 | There'll be more than enough to do just to get through the basic parts of the assignments and of course on the forums

00:18:11.060 | We'd be very happy to help you with all of your questions there

00:18:13.960 | And if you are more advanced we really appreciate your help in adding new material to the wiki

00:18:19.020 | Answering others questions on the forums people started their own threads on the forums around kind of outside related topics that they were interested in

00:18:26.880 | There are a lot of different ways to be involved

00:18:29.780 | So here's a couple of quotes we got from people after they completed the in-person course

00:18:34.700 | And this is one that we heard again and again so for example

00:18:38.500 | this person says I personally fell into the habit of watching the lectures too much and

00:18:43.540 | Googling definitions and concepts and so forth too much without running the code at first

00:18:49.180 | I thought that I should read the code quickly and then spend time researching the theory behind it

00:18:53.900 | In retrospect I should have spent the majority of my time on the actual code in the notebooks instead in terms of running it

00:19:01.020 | And seeing what goes into it and seeing what comes out of it and Rachel

00:19:04.900 | I know you've seen similar things in your past teaching experience

00:19:07.760 | I've seen this in teaching full stack software development and test students

00:19:11.940 | And I also know that I've been guilty a bit myself sometimes

00:19:15.140 | And that was that students would sometimes kind of rather than start their project

00:19:19.980 | They would keep doing more and more research reading more and more tutorials and feeling like there's more and more they need to learn before

00:19:25.220 | They can start coding and two problems with that are one and I mean you want to have some background before you begin

00:19:31.900 | But there's a point where you just need to start coding

00:19:33.900 | Because you can't know exactly what you're going to need until you start

00:19:37.300 | Coding and building and seeing what errors you get and what things you don't know how to do

00:19:41.740 | And then secondly the test of whether you understand something is whether you can build with it and so kind of reading tutorials

00:19:49.780 | It's very possible to think oh, I understand all this, but it's not till you're writing code yourself

00:19:55.020 | I'm kind of seeing what your what your error rates are and what what's working and what's not that you know whether or not you truly

00:20:00.940 | Understand something yeah, so when I saw students at the study sessions during the week at USF

00:20:06.860 | I would keep telling them the same thing again

00:20:09.100 | And again just don't stop and wait till you feel ready to code start coding now

00:20:13.300 | And it's through that coding experience that you're actually going to figure out what you don't know and what you do know and you'll be

00:20:19.460 | able to develop the intuition by running lots of experiments

00:20:23.500 | This is another interesting quote

00:20:25.500 | from somebody talking about

00:20:27.980 | This learning style he said it's been very interesting learning from somebody who is an entrepreneur

00:20:33.060 | That'd be me a very known nonsense approach to getting things done very hands-on very smart and driven

00:20:39.900 | Your usual career and structure is quite the opposite

00:20:42.900 | So it's been refreshing and even somewhat shocking this is possibly understating things a bit it can be in fact

00:20:49.860 | We heard from quite a few people at the start. It was somewhat shocking to find so many things

00:20:55.140 | taught so quickly and it kind of

00:20:59.500 | Can seem like such a high level, but of course by the end of the seven weeks and assuming that each time you're putting 10

00:21:06.740 | Hours into those weeks you've actually got

00:21:09.700 | Many many full end-to-end processes under your belt so by the end of it you actually are going to develop a very deep and complete understanding

00:21:19.500 | Yeah, I know after the first lesson

00:21:21.060 | I heard a number of students kind of say things like oh

00:21:23.420 | I I didn't really get the details from that lesson and you know

00:21:26.860 | I feel like I need to spend all this time understand studying the details

00:21:29.700 | And we hadn't taught the details in the first lesson

00:21:32.380 | And the idea is that kind of we went more and more in depth each time

00:21:36.540 | You're seeing this end-to-end process and then kind of as time goes on digging into it more

00:21:41.740 | But even after the first lesson you can apply it you can actually create

00:21:46.500 | World-class image recognition models, and so you can go back to your organization and start trying things

00:21:52.180 | This is something else we encourage people to do try things with your own data and your own problems from the very first lesson

00:21:58.340 | So it's been a interesting experience in every way even the way we built this course was unusual

00:22:05.440 | For example, I actually wrote most of the material while traveling from the northern tip to the southern tip of Japan. I

00:22:14.700 | coded and wrote in every possible place you can imagine

00:22:19.260 | And this was a really an experiment for me because I studied human learning theory a lot

00:22:25.180 | And I know that in theory human creativity is meant to be better when you have a wider variety of contexts

00:22:32.180 | Interestingly, I actually found I was more productive in that month than I feel I ever have been before and you'll see actually in the

00:22:40.380 | material that you learn we show a lot of

00:22:44.780 | new techniques or different techniques or different ways of thinking about things and I think this kind of

00:22:49.860 | Different way of building the course perhaps what was really helpful and coming up with this kind of more creative approach

00:22:56.140 | So not everybody in the course

00:23:01.700 | In the in the in person course were able to put in the at least eight hours a week, but the vast majority were

00:23:10.060 | and those who didn't

00:23:14.260 | Still completed the course

00:23:15.900 | They just found they didn't necessarily pick everything up the way that they hoped they would

00:23:20.540 | But of course the nice thing is you can always come back to it later

00:23:23.580 | So our suggestion would be now that it's a MOOC now that you don't have to do it every single week

00:23:30.060 | Ideally you will put in the 10 hours a week. Did you want to talk about those 10 hours a little bit Rachel?

00:23:35.860 | Yeah, we wanted to give you kind of some suggestions on how how to use that time

00:23:40.660 | So the videos are between two and two and a half hours long

00:23:43.980 | and so

00:23:47.100 | with those videos

00:23:49.260 | You may find it helpful as you review them to use these notes

00:23:53.420 | Yeah, so this is coming from our wiki with you.fast.ai. There's a page for each lesson that has

00:23:59.900 | notes

00:24:01.700 | Kind of about the lesson. It also has links to other other resources

00:24:06.540 | That may be useful to you

00:24:09.340 | These notes are pretty complete. They're not designed to be read entirely independently from the video lesson

00:24:15.380 | But they are something which you can

00:24:17.620 | Read on your way to work. Maybe when you don't have

00:24:20.660 | It's not convenient to actually watch the lesson

00:24:24.060 | Sorry, I was gonna say we're expecting that you'll watch the lessons more than once, you know

00:24:29.740 | So the first time through you're kind of watching to get maybe a lot of the high-level ideas

00:24:33.980 | Then you'll probably want to read the wiki try out the notebooks and then go back and watch the lesson again

00:24:40.600 | Kind of maybe to get more detail. Yeah, I don't think any of our students in the in-person course just watch the lessons once

00:24:46.940 | They saw them live of course

00:24:48.660 | But then we also they also had the recording from the next day and I think everybody has spoken to watch them at least twice

00:24:55.020 | And then of course the other thing you've got is the notebooks the notebooks as you see have quite a lot of pros in them

00:25:01.620 | as well

00:25:03.100 | They've got quite a lot of additional detail that we don't necessarily get into in the video lesson

00:25:08.840 | But most importantly as we described they give you an environment in which you can experiment

00:25:13.780 | In fact, not only do we suggest that you experiment we have a very specific suggestion about how to use these notebooks

00:25:22.080 | Yeah, so we recommend that you

00:25:24.740 | Read through the notebook and then and this is after you've watched the video

00:25:28.540 | At least once and everything makes sense put it aside and try creating a new a new notebook where you go through that process yourself

00:25:36.340 | And so this is from scratch, right? This is like creating your own notebook to test that you can actually build it yourself

00:25:43.180 | Yeah

00:25:43.460 | We do not want you to just hit shift enter shift enter and run through the existing notebook

00:25:47.940 | Because again the test of whether whether you know something is can you can you build and code with it yourself?

00:25:54.020 | So if you get stuck you can always then go back and refer to the class notebook and then rather than copying and pasting it

00:26:00.300 | Make sure you do understand what it's saying

00:26:02.900 | Maybe look up some documentation about that concept and then put that notebook aside and see if you can now do it yourself

00:26:09.540 | So in a sense you're plagiarizing a lot from the notebook, but you're plagiarizing in a good way

00:26:15.180 | You know you're plagiarizing not by copying and pasting but by plagiarizing the concepts and making sure that you can recreate them yourself

00:26:23.440 | And then if you have a have questions

00:26:26.000 | Please ask them the forums are the first place you should go to and first search to see if someone's already asked your question

00:26:32.800 | As we said earlier, there's a separate thread for each lesson that are already

00:26:36.760 | Have tons of helpful questions and answers from the students that took our in-person course

00:26:41.960 | In fact, there's a great quote which we talk about in one of the lessons from the head of Google Brain

00:26:46.600 | Who says that their rule at Google Brain is that if you have a problem you first of all try to fix it yourself for half

00:26:53.360 | an hour

00:26:54.760 | And if half a half an hour you can't fix it yourself you then have to ask somebody so that ensures that you

00:27:01.680 | Always give it a go yourself and hopefully learn from the experience

00:27:06.520 | But you never waste too much time on something which somebody else can help you with. Yes, it's great advice

00:27:13.880 | So as Rachel said the forums are a really helpful resource and when you go to the forums

00:27:18.540 | You'll find that there's a lot of existing discussions

00:27:21.440 | There's a separate discussion for every lesson for example and each and each of those discussions

00:27:26.720 | You'll see that there's a summary of the existing discussion at the start

00:27:29.880 | So you may find that what you need is already in the question and answers there

00:27:35.080 | If it's not of course

00:27:36.680 | Feel free to add your question and you're generally found it's responded to within a small number of hours

00:27:42.840 | Maybe by Rachel or I or maybe by one of the other students

00:27:46.160 | The other thing you may find helpful is that each lesson has a timeline on the wiki and those hyperlinks are actually hyperlinks

00:27:54.840 | Directly to the part of the video which discusses that topic

00:27:57.960 | So if you're trying to remember how momentum works

00:28:00.960 | You can just click on that link and you'll jump straight to me telling you about momentum

00:28:05.240 | As Rachel said there's also a number of resources available to help you

00:28:12.360 | So this is taken from the front page of our wiki

00:28:15.660 | There's a whole section of tools with links where you can learn about learn more about each of the pieces that we use in the development

00:28:21.820 | Environment and so our goal here is not to be a single source of truth

00:28:25.800 | If somebody else has already done a great job of teaching one of these tools

00:28:29.040 | We'll leave it to them

00:28:30.680 | So we don't attempt to give you a great bash reference or a great umpire reference because people have already done that

00:28:37.880 | So if you want to learn more about one of these things jump onto the wiki click through here

00:28:42.800 | And you'll find some curated resources that we think are really helpful

00:28:46.220 | So this lesson

00:28:49.840 | You're going to cover a lot of stuff

00:28:52.240 | But these are the four things to keep in mind

00:28:54.440 | By the end of the lesson you want to make sure that you can create an AWS instance

00:28:58.800 | That you can connect to it with SSH

00:29:01.320 | It can run a Jupyter notebook in it and you can run those that state-of-the-art custom model code that we showed you earlier

00:29:08.480 | Those first three things you're going to be doing every single project in every single lesson

00:29:14.040 | So you're going to want to be really comfortable at doing that and for those of you who don't I haven't done that before

00:29:19.040 | It might take you a little while to get the hang of it and maybe a few

00:29:22.520 | unsuccessful attempts first

00:29:25.440 | So this first lessons unusual and that it's a lot more about

00:29:29.320 | Kind of getting your development environment set up and not as much about deep learning

00:29:33.560 | So indeed if you've got a background in Python and AWS and Linux

00:29:39.240 | You may find this lesson on the easy side in which case you can dip through it pretty fast

00:29:43.400 | If you don't have a background in these tools today's class may seem really overwhelming

00:29:48.140 | And we don't don't want you to be discouraged by that because this is very different from the future lessons

00:29:53.980 | But it's necessary to get your environment set up so that you can be coding throughout the course

00:29:58.120 | Yeah, I mean the folks who didn't have that background in the in-person course once they actually got through this and

00:30:04.080 | Often it was a lot of work and it was pretty tough

00:30:06.720 | But at the end they finally got the point and they could say, okay

00:30:09.600 | I've set up a GPU instance in the cloud

00:30:11.760 | I've set up my development environment and I have trained from scratch a model that can recognize dogs from cats

00:30:18.120 | And it was very very exciting. So if this is hard work for you

00:30:22.240 | Just know that when you get through the other end of it, it's going to be really exciting

00:30:27.120 | So I asked that everyone who's trying to decide if this course is for them

00:30:30.480 | Try at least the first two lessons since the first lesson is so much about setup

00:30:36.040 | Yeah, these lessons become

00:30:38.640 | Obviously we build more and more on the techniques we've learned and so we're going to be using this infrastructure in every lesson

00:30:45.080 | By the time we get to lesson seven, we're going to be looking at some pretty sophisticated and custom neural network architectures

00:30:52.560 | We're going to cover every different type of SGD optimization

00:30:57.040 | We're going to be covering convolutional neural networks and recurrent neural networks. So there's going to be a lot of exciting stuff

00:31:02.360 | and yeah, we really look forward to seeing you on the forums and

00:31:06.680 | Good luck with learning about deep learning

Lesson 1 Overview

Chapters