1 00:00:02,100 --> 00:00:04,350 Everyone, and welcome back to this class. 2 00:00:08,310 --> 00:00:14,100 This lecture is designed to answer some common questions I get about how difficult this course is and 3 00:00:14,100 --> 00:00:19,280 where it fits in terms of academia, the professional world and its practical applicability. 4 00:00:20,010 --> 00:00:25,530 So rather than answering this question individually over and over again, it seems like a better idea 5 00:00:25,530 --> 00:00:27,750 to make a lecture about it and answer everybody. 6 00:00:27,750 --> 00:00:33,630 At the same time, a lot of people become confused because this is their first machine learning course 7 00:00:33,630 --> 00:00:36,840 and they just don't know where it fits in the grand scheme of things. 8 00:00:37,320 --> 00:00:41,880 So is this course academic or practical and is it for beginners or experts? 9 00:00:42,120 --> 00:00:43,950 Is it fast paced or slow paced? 10 00:00:49,020 --> 00:00:53,970 I'm going to answer the second question first, is this cause for beginners or experts? 11 00:00:55,130 --> 00:01:01,250 Interestingly, the problem with this question is a problem for machine learning to let me explain why 12 00:01:01,250 --> 00:01:03,170 this question is itself problematic. 13 00:01:03,940 --> 00:01:09,870 One of the challenges in natural language processing is that of ambiguity, when someone says some words, 14 00:01:09,880 --> 00:01:15,220 the problem is those words can mean multiple things and it just so happens that I have some courses 15 00:01:15,220 --> 00:01:15,740 on it up. 16 00:01:15,760 --> 00:01:18,020 So this is a problem I think about often. 17 00:01:18,550 --> 00:01:19,950 So how does that apply here? 18 00:01:20,890 --> 00:01:26,170 Well, when we use the words beginner and expert, we have to ask a beginner from whose perspective 19 00:01:26,320 --> 00:01:31,920 an expert from whose perspective for some of you, you might think that a beginner is a random person. 20 00:01:31,930 --> 00:01:33,700 You take off the street who knows nothing. 21 00:01:34,180 --> 00:01:35,660 But even that is ambiguous. 22 00:01:36,550 --> 00:01:41,110 There are people out there who do not even really know how to use a computer beyond turning it on and 23 00:01:41,110 --> 00:01:42,800 checking their email and browsing the web. 24 00:01:43,510 --> 00:01:48,550 There are people out there who are pros at Microsoft Office so they know how to use a computer and make 25 00:01:48,550 --> 00:01:53,110 a Microsoft Word document or a spreadsheet, but they've never done any programming before. 26 00:01:55,020 --> 00:01:59,520 There are people out there who are very good developers and have lots of programming experience, but 27 00:01:59,550 --> 00:02:01,320 they have no idea what machine learning is. 28 00:02:02,430 --> 00:02:07,470 So all of these people are beginners in some way, but each of them would require a vastly different 29 00:02:07,470 --> 00:02:09,930 amount of work to prepare for this course. 30 00:02:14,910 --> 00:02:21,500 It's the same situation with experts most people consider anyone who knows more than them to be an expert, 31 00:02:22,080 --> 00:02:26,100 but of course, how much you know and how much some other person knows is relative. 32 00:02:26,940 --> 00:02:32,040 If you know how to plug and play into you learn and nothing else, you might seem like an expert to 33 00:02:32,040 --> 00:02:37,620 someone who only knows Excel, but you'll seem like a beginner to someone who's read Bishop's pattern 34 00:02:37,620 --> 00:02:39,270 recognition and machine learning book. 35 00:02:44,330 --> 00:02:49,020 So just to finalize our answer to this question, is this cause for beginners or experts? 36 00:02:49,580 --> 00:02:55,080 This question is misinformed to begin with because beginners and experts are not well-defined terms. 37 00:02:55,850 --> 00:03:01,900 Instead, we have to go back to what is the specific skill set you need before starting this course. 38 00:03:02,420 --> 00:03:06,880 And of course, that's defined by the prerequisites which are listed in the course description. 39 00:03:07,370 --> 00:03:09,050 So precision is the key here. 40 00:03:09,500 --> 00:03:15,200 We want to know what is the precise list of skills that I need before taking this course. 41 00:03:15,710 --> 00:03:18,950 This is much better than ambiguous terms like beginner and expert. 42 00:03:23,860 --> 00:03:28,990 Another matter is that people are going to measure their own skills differently, that's a much tougher 43 00:03:28,990 --> 00:03:29,950 issue to deal with. 44 00:03:30,460 --> 00:03:35,800 Sure, I can say you need to know python coding before starting this course, but you can be sure that 45 00:03:35,800 --> 00:03:41,230 there's a big difference between someone who just learned hello world and for loops versus someone working 46 00:03:41,230 --> 00:03:42,820 at a big financial institution. 47 00:03:47,820 --> 00:03:53,160 Now, even in those big financial institutions, there could be big gaps in performance, even within 48 00:03:53,160 --> 00:03:55,580 your own team, there can be big gaps in performance. 49 00:03:56,130 --> 00:04:01,080 I've seen guys who've been working for 15 years do worse than guys who've been working less than one 50 00:04:01,080 --> 00:04:01,390 year. 51 00:04:02,070 --> 00:04:08,190 So even if I say you need to know python coding, there can be a lot of variation in skill among those 52 00:04:08,190 --> 00:04:09,600 who consider themselves good at it. 53 00:04:10,340 --> 00:04:12,750 A similar situation happens with deep learning. 54 00:04:13,080 --> 00:04:17,670 I can say you need to know how neural networks work, but there's a big difference between someone who 55 00:04:17,670 --> 00:04:22,710 can code them from scratch and has experience using them versus someone who just watched them online 56 00:04:22,710 --> 00:04:24,630 animations for a few hours. 57 00:04:25,230 --> 00:04:27,610 If you had to bet on who knows neural networks better. 58 00:04:27,900 --> 00:04:28,770 Who would you pick? 59 00:04:29,790 --> 00:04:34,740 So it's a really answer this question, we have to turn to the course itself, if I say you need to 60 00:04:34,740 --> 00:04:40,170 know python coding and you claim that you do, but you don't understand some python code in this course, 61 00:04:40,860 --> 00:04:44,760 then simply put, your python coding skills are not yet up to par. 62 00:04:45,160 --> 00:04:51,000 So you have to improve your python coding skills by asking questions on the Q&A and taking the appropriate 63 00:04:51,000 --> 00:04:52,730 steps to cover your gaps. 64 00:04:57,910 --> 00:05:03,940 Quite often I get people who are just too confident in their own abilities when unfortunately the reality 65 00:05:03,940 --> 00:05:09,100 is there is a huge discrepancy between what they think they know and what they actually know. 66 00:05:09,970 --> 00:05:14,020 What's funny is these people sometimes say this course requires PhD math. 67 00:05:14,590 --> 00:05:21,520 I assure you there's absolutely no math in any of my courses that requires a Ph.D. or that only a Ph.D. 68 00:05:21,530 --> 00:05:22,420 can understand. 69 00:05:22,840 --> 00:05:26,110 My courses contain only undergraduate math at most. 70 00:05:26,860 --> 00:05:32,170 Sometimes people even claim to have a Ph.D. and then claim that it's still too hard, which is funny 71 00:05:32,170 --> 00:05:38,560 because that essentially invalidates your entire Ph.D. The entire point of a PhD is to turn you into 72 00:05:38,560 --> 00:05:40,060 an independent researcher. 73 00:05:40,330 --> 00:05:45,040 And so if you're supposed to be at the level of doing independent research, but you can't understand 74 00:05:45,040 --> 00:05:47,830 some undergraduate math, well, that's not very good. 75 00:05:52,930 --> 00:05:58,540 Essentially, what it boils down to is people are really trying to say everything I don't understand 76 00:05:58,540 --> 00:05:59,330 is not important. 77 00:05:59,800 --> 00:06:02,020 Everything I do understand is important. 78 00:06:02,290 --> 00:06:06,250 And therefore, everything you say that I don't understand, I don't need to know. 79 00:06:06,890 --> 00:06:09,010 You can see why this thinking is dangerous. 80 00:06:09,460 --> 00:06:14,470 If you truly believe that everything you do not know is not important, then you'll never learn anything 81 00:06:14,470 --> 00:06:14,800 new. 82 00:06:19,800 --> 00:06:24,330 Now, this brings us to our third question, is this course fast paced or slow paced? 83 00:06:24,900 --> 00:06:26,850 And I think you can see where I'm going with this. 84 00:06:27,360 --> 00:06:29,610 This is yet another example of ambiguity. 85 00:06:30,150 --> 00:06:35,890 Fast pace to whom, whether it's fast paced or not, depends on which person is taking the course. 86 00:06:36,420 --> 00:06:39,800 So if someone is well prepared, then things will be just right. 87 00:06:40,470 --> 00:06:44,550 If someone already knows most of these topics, then it will probably be too slow. 88 00:06:45,480 --> 00:06:50,940 If someone chose not to prepare for the course or they have an inadequate understanding of the prerequisites, 89 00:06:51,270 --> 00:06:52,810 then it's going to be too fast. 90 00:06:53,460 --> 00:06:58,110 So if I expected you to know something and you didn't, then you have to look it up, which is going 91 00:06:58,110 --> 00:07:00,270 to make the course feel like it's going too fast. 92 00:07:05,290 --> 00:07:11,620 So here is the real test, if I say something depends on calculus of probability and you claim to know 93 00:07:11,830 --> 00:07:17,070 calculus and probability, but you don't understand the thing that depends on calculus and probability, 94 00:07:17,380 --> 00:07:23,800 you have to ask yourself, honestly, do I really understand calculus and probability or did the instructor 95 00:07:24,040 --> 00:07:27,700 invent his own kind of calculus and probability that I don't yet understand? 96 00:07:28,180 --> 00:07:29,920 Well, of course, that's not very likely. 97 00:07:30,430 --> 00:07:35,320 It's most likely that you thought you understood calculus and probability, but you really don't. 98 00:07:40,350 --> 00:07:44,040 Now on to our initial question, what kind, of course, is this? 99 00:07:44,070 --> 00:07:45,060 Is it academic? 100 00:07:45,390 --> 00:07:46,650 Is it for professionals? 101 00:07:46,650 --> 00:07:49,340 Is a practical sort of understand this. 102 00:07:49,350 --> 00:07:53,970 You have to understand where machine learning is usually taught in how it is usually taught. 103 00:07:54,980 --> 00:08:00,110 Firstly, machine learning is usually taught in upper year computer science and engineering programs 104 00:08:00,260 --> 00:08:01,850 in university and college. 105 00:08:02,360 --> 00:08:07,370 So these are students who already took calculus one to three linear algebra, differential equations, 106 00:08:07,550 --> 00:08:13,700 probability and statistics, discrete math programming and possibly more all in their first two or three 107 00:08:13,700 --> 00:08:14,580 years of college. 108 00:08:15,140 --> 00:08:19,610 So by the time they get to third and fourth year, they know this stuff and they're ready to apply it. 109 00:08:24,690 --> 00:08:30,420 So a typical academic course might cover a whole library of machine learning algorithms, so you might 110 00:08:30,420 --> 00:08:36,840 learn K nearest neighbor, logistic regression K means clustering PCA and so on, all in the same course. 111 00:08:37,870 --> 00:08:43,090 Typically, you'd cover one algorithm per lecture, which is about two hours overall, you might have 112 00:08:43,090 --> 00:08:45,960 something like 12 or maybe 20 of those lectures in a term. 113 00:08:46,660 --> 00:08:50,920 And by cover, I mean mostly via math, geometry and derivations. 114 00:08:51,430 --> 00:08:55,150 These courses don't typically have you at your computer doing programming work. 115 00:08:55,870 --> 00:09:00,850 If you do end up doing programming, it's usually part of a lab assignment, which is actually just 116 00:09:00,850 --> 00:09:03,830 a very small part of the course compared to the rest. 117 00:09:04,510 --> 00:09:10,030 So it's mostly theoretical dealing with equations, solving equations, reasoning about equations. 118 00:09:10,480 --> 00:09:14,220 Those equations typically involve probabilities and information theory. 119 00:09:14,680 --> 00:09:18,160 So you have two hours to go through the theory for one algorithm. 120 00:09:18,160 --> 00:09:22,860 And then if there's coding involved, which there usually isn't, you do it on your own time. 121 00:09:23,530 --> 00:09:26,170 The lectures themselves do not involve any coding. 122 00:09:31,160 --> 00:09:34,970 But there's an even bigger problem with these academic courses and what is that? 123 00:09:35,870 --> 00:09:38,540 What happens is that the grades turn out to be very low. 124 00:09:41,030 --> 00:09:45,650 Out of 100 percent, the average might be around 25 percent or up to 40 percent. 125 00:09:46,190 --> 00:09:48,460 Now you might think, wow, so everyone fails. 126 00:09:48,860 --> 00:09:53,280 And of course, that's not the case since everyone gets shifted based on the average. 127 00:09:53,870 --> 00:09:55,490 So what is the real problem here? 128 00:09:56,090 --> 00:10:01,040 A lot of people ask me, well, why don't you cover every machine learning algorithm in one big mass, 129 00:10:01,040 --> 00:10:03,230 of course, and just call it machine learning? 130 00:10:04,040 --> 00:10:09,860 Well, the answer is that if you tried to do too much, you only end up understanding 25 percent of 131 00:10:09,860 --> 00:10:12,670 the material, which I think is clearly demonstrated here. 132 00:10:13,460 --> 00:10:15,550 And as they say, the proof is in the pudding. 133 00:10:16,010 --> 00:10:20,780 If you look at these courses where students are trying to digest 10 different algorithms at the same 134 00:10:20,780 --> 00:10:22,410 time, what is the result? 135 00:10:22,970 --> 00:10:24,990 The result is they don't understand any of them. 136 00:10:25,370 --> 00:10:29,150 If you try to understand too many things, you end up understanding nothing. 137 00:10:30,020 --> 00:10:35,780 I take the perspective that if you're going to take a course, you should aim to understand 100 percent 138 00:10:35,780 --> 00:10:36,650 of the material. 139 00:10:37,880 --> 00:10:43,490 Each course is broken up into appropriate levels based on how difficult they are and by topic. 140 00:10:43,940 --> 00:10:50,030 So, for example, you can typically take Calculus One, which focuses on differential calculus, Calculus 141 00:10:50,030 --> 00:10:55,890 two, which focuses on integral calculus and calculus three, which focuses on multivariable calculus. 142 00:10:56,420 --> 00:11:01,340 You would never learn all of these and just one huge course called math, because that's just too much 143 00:11:01,340 --> 00:11:03,410 stuff to understand all at once. 144 00:11:04,040 --> 00:11:10,100 I want you to understand, one hundred percent of the material, even 50 percent or 75 percent, is 145 00:11:10,100 --> 00:11:11,600 not good enough in my opinion. 146 00:11:12,140 --> 00:11:18,470 If you only understand 75 percent of the material and the next course depends on the 25 percent that 147 00:11:18,470 --> 00:11:22,160 you didn't understand, well, then you're not going to make it to the next level. 148 00:11:27,200 --> 00:11:32,360 Now, let's look at the opposite end of the spectrum, what most people might consider to be a practical 149 00:11:32,360 --> 00:11:32,810 course. 150 00:11:33,750 --> 00:11:38,790 I don't consider these to be practical courses, and I'll explain why later, but I use the word practical 151 00:11:38,790 --> 00:11:43,170 here in quotes, since that's what some of the more beginner students will refer to the mess. 152 00:11:43,980 --> 00:11:46,440 So how does a practical course typically proceed? 153 00:11:48,000 --> 00:11:52,950 Well, first, you were never really taught how the algorithms work other than at a very high level. 154 00:11:53,520 --> 00:11:56,890 You might learn how they work by way of analogies and metaphors. 155 00:11:57,330 --> 00:12:00,610 So, for example, suppose you're learning about gradient descent. 156 00:12:01,200 --> 00:12:04,560 Well, the instructor might relate this to a ball rolling down a hill. 157 00:12:05,820 --> 00:12:10,680 Now, some beginner students, they might conclude that I understand a ball rolling down a hill. 158 00:12:10,860 --> 00:12:12,960 Therefore, I understand gradient descent. 159 00:12:13,500 --> 00:12:16,050 But they neglect to understand how to take derivatives. 160 00:12:16,260 --> 00:12:18,690 They neglect to understand how to choose a good learning rate. 161 00:12:19,020 --> 00:12:22,650 And most importantly, they neglect how to implement gradient descent in code. 162 00:12:24,130 --> 00:12:29,230 So it's totally wrong to assume that just because you can visualize a ball rolling down a hill, that 163 00:12:29,230 --> 00:12:31,720 you understand gradient descent, why? 164 00:12:32,840 --> 00:12:36,210 Well, the problem is everybody can imagine a ball rolling down a hill. 165 00:12:36,650 --> 00:12:41,600 Your grandparents, who can't even turn on a computer, can understand a ball rolling down a hill. 166 00:12:41,990 --> 00:12:46,600 So don't equate understanding an analogy to understanding the real topic. 167 00:12:51,760 --> 00:12:57,580 So what's the next thing these practical courses do, the next thing they usually teach you is how to 168 00:12:57,580 --> 00:13:02,420 plug into an API or in other words, make use of a library, someone else wrote. 169 00:13:02,980 --> 00:13:04,570 This is also not ideal. 170 00:13:05,230 --> 00:13:07,830 Typically, this involves just two or three lines of code. 171 00:13:08,410 --> 00:13:13,100 You have to wonder if machine learning just boils down to two or three lines of code. 172 00:13:13,450 --> 00:13:15,200 Why is it such a big deal these days? 173 00:13:15,610 --> 00:13:18,640 Why did it take decades for people to realize its usefulness? 174 00:13:19,210 --> 00:13:23,500 And of course, the answer is that there's much more to machine learning than just these two or three 175 00:13:23,500 --> 00:13:24,220 lines of code. 176 00:13:24,880 --> 00:13:29,140 When you use a deep learning library like Carus, you're not doing deep learning. 177 00:13:29,470 --> 00:13:31,720 You were using a deep learning interface. 178 00:13:31,730 --> 00:13:33,010 You're simply programming. 179 00:13:33,400 --> 00:13:39,610 It's more accurate to say I can program with deep learning libraries rather than saying I actually know 180 00:13:39,610 --> 00:13:40,210 deep learning. 181 00:13:44,860 --> 00:13:47,500 But some of you might say, well, it's all I really want to do. 182 00:13:47,710 --> 00:13:53,070 All I need is an API and for the API to give me an answer, I don't care about how it works. 183 00:13:53,560 --> 00:13:55,460 But then here is where you run into trouble. 184 00:13:56,140 --> 00:13:58,780 Recall that in addition to all data is the same. 185 00:13:59,110 --> 00:14:01,930 We know that all machine learning interfaces are the same. 186 00:14:02,470 --> 00:14:05,520 So the question becomes, which API do you choose? 187 00:14:05,530 --> 00:14:10,420 If they're all the same, how can you make an informed choice about which one to use? 188 00:14:11,110 --> 00:14:15,920 Well, the reality is, if you don't know how these things work, then you can't answer this question. 189 00:14:16,540 --> 00:14:21,840 This is why some of these so-called practical courses can cover 20 algorithms in the same course. 190 00:14:22,120 --> 00:14:27,760 It's because they never actually talk about how the algorithms work since the API is the same every 191 00:14:27,760 --> 00:14:28,130 time. 192 00:14:28,270 --> 00:14:32,620 It's very easy to just do the same thing 20 times and then not bother with the details. 193 00:14:33,310 --> 00:14:37,570 So this gives people a lot of confidence that they were able to do the same thing 20 times. 194 00:14:38,050 --> 00:14:42,760 But it's dangerous thinking because you don't realize you really only learned one thing. 195 00:14:42,760 --> 00:14:44,500 You just repeated it 20 times. 196 00:14:49,500 --> 00:14:52,050 So how about when it comes to the professional world? 197 00:14:52,930 --> 00:14:58,420 Sometimes people tend to believe that in your profession, you'll be leaning more toward the practical 198 00:14:58,420 --> 00:15:02,860 end of the spectrum where you don't really need to know how a machine learning model works. 199 00:15:03,040 --> 00:15:04,570 You just have to know its API. 200 00:15:05,500 --> 00:15:08,750 But what it really comes down to is what your job actually is. 201 00:15:09,190 --> 00:15:13,900 If you're a regular everyday programmer and you just want to see how a machine learning model might 202 00:15:13,900 --> 00:15:17,380 perform on your data, then you're going to use a pre-built library. 203 00:15:17,380 --> 00:15:21,250 Most likely that's just the two or three lines of code I referenced earlier. 204 00:15:22,750 --> 00:15:28,210 However, that does not mean you don't still have to conform to best practices, make sure you're not 205 00:15:28,210 --> 00:15:30,070 mixing your test set with your transfer. 206 00:15:30,460 --> 00:15:33,070 Make sure you're doing proper validation and so on. 207 00:15:33,520 --> 00:15:38,500 And when it comes to tuning your model, then you have to be able to dig down into the details and see 208 00:15:38,500 --> 00:15:42,840 which hyper parameters are possible to change and which are likely to have the most impact. 209 00:15:43,780 --> 00:15:48,270 So only knowing how to use an API is only going to get you past the first stage. 210 00:15:48,640 --> 00:15:52,630 You can plug in your data and at least your code is the right syntax and doesn't crash. 211 00:15:52,960 --> 00:15:56,740 But of course, your code not crashing doesn't necessarily mean it's right. 212 00:16:01,710 --> 00:16:06,660 Back on the other end of the spectrum, you have programmers and scientists whose sole job it is to 213 00:16:06,660 --> 00:16:07,560 do machine learning. 214 00:16:08,070 --> 00:16:09,950 These guys are machine learning experts. 215 00:16:10,350 --> 00:16:14,090 They know the math inside and out and have a full computer science education. 216 00:16:14,970 --> 00:16:20,400 And so to even be able to communicate your ideas effectively, you have to be comfortable with a theory 217 00:16:20,400 --> 00:16:23,430 and not just how to plug into an API with two lines of code. 218 00:16:24,330 --> 00:16:29,910 These scientists are building custom models, really digging down into the math to not only use but 219 00:16:29,910 --> 00:16:33,420 invent new things and to be able to invent new things. 220 00:16:33,600 --> 00:16:35,600 You really need to know the math very well. 221 00:16:36,180 --> 00:16:40,290 And so if you're on a team like this, you can't just know how to write two lines of code. 222 00:16:40,560 --> 00:16:43,050 Your skill set has to be far beyond that. 223 00:16:43,950 --> 00:16:49,290 So your theoretical background has to be very solid if you want to be able to communicate effectively 224 00:16:49,410 --> 00:16:50,610 with other scientists. 225 00:16:55,740 --> 00:16:59,880 Now, a lot of people say, well, I don't ever want to be a scientist, so I don't really care how 226 00:16:59,880 --> 00:17:05,220 these algorithms work, unfortunately, that's a very bad attitude to have in the workplace. 227 00:17:05,220 --> 00:17:11,310 I suppose, for example, I am running a business now, an employee of mine, he wants to show off and 228 00:17:11,310 --> 00:17:13,170 say he's learning about machine learning. 229 00:17:13,650 --> 00:17:18,450 He says, oh, I'm taking this machine learning course online, therefore I will become a machine learning 230 00:17:18,450 --> 00:17:18,980 master. 231 00:17:19,620 --> 00:17:23,050 But his approach to machine learning is to learn as little as possible. 232 00:17:23,430 --> 00:17:25,080 He says he doesn't care about the math. 233 00:17:25,080 --> 00:17:26,610 He only wants the practical stuff. 234 00:17:26,970 --> 00:17:29,000 He only wants to know the bare minimum. 235 00:17:29,640 --> 00:17:31,560 He tells me he doesn't care how it works. 236 00:17:31,560 --> 00:17:36,210 He's just going to plug into some API because that's what his buddy told him he could do. 237 00:17:37,240 --> 00:17:43,420 As a business owner, I really don't want this guy working for me, I run a business that sells something 238 00:17:43,750 --> 00:17:49,810 and the life of the company itself and the livelihoods of the employees all depend on our products being 239 00:17:49,810 --> 00:17:51,430 the best they can possibly be. 240 00:17:51,940 --> 00:17:56,560 So if someone comes to me and says, I don't care about being the best, I just care about doing the 241 00:17:56,560 --> 00:18:00,140 bare minimum, my response is I don't want that guy on my team. 242 00:18:00,460 --> 00:18:01,960 In fact, he should be fired. 243 00:18:02,560 --> 00:18:05,090 So be careful if this is your approach to machine learning. 244 00:18:05,590 --> 00:18:10,720 No business owner wants to hear that your approach to the business is to do the bare minimum and to 245 00:18:10,720 --> 00:18:12,980 put off some math because you think it's too hard. 246 00:18:13,450 --> 00:18:15,610 That's a scary prospect for a business owner. 247 00:18:20,610 --> 00:18:25,050 Now, if we go back to our original question, maybe you've realized we haven't answered it yet. 248 00:18:25,650 --> 00:18:30,800 What kind, of course, is this where you can think of it as a marriage between these approaches? 249 00:18:31,470 --> 00:18:35,330 We want to cover all the math and derivations because that's very important. 250 00:18:35,640 --> 00:18:40,560 If you want to be able to communicate effectively with other scientists, to be able to debug your code 251 00:18:40,560 --> 00:18:43,590 effectively and to be able to invent new approaches. 252 00:18:44,310 --> 00:18:46,620 This is also what we mean by machine learning. 253 00:18:47,100 --> 00:18:51,200 We mean really knowing machine learning and not just programming with an API. 254 00:18:51,990 --> 00:18:54,360 So the theoretical background is very important. 255 00:18:54,810 --> 00:19:00,630 At the same time, we're not going to introduce 20 different algorithms in the same course because unlike 256 00:19:00,630 --> 00:19:06,870 a typical academic course where you can pass just by understanding 25 percent of the material in this 257 00:19:06,870 --> 00:19:12,120 course, it's structured in a way that if you have any missing gaps, it's going to be impossible for 258 00:19:12,120 --> 00:19:13,710 you to move on to the next stage. 259 00:19:15,080 --> 00:19:18,150 But academic courses don't contain much coding, if at all. 260 00:19:18,620 --> 00:19:22,880 So in this course, every algorithm we learn about will be implemented in code. 261 00:19:23,450 --> 00:19:27,090 My motto is, if you can't implement it, then you don't understand it. 262 00:19:27,830 --> 00:19:29,390 You can pretend to understand it. 263 00:19:29,600 --> 00:19:32,260 And when you talk about it, you might sound like you understand it. 264 00:19:32,510 --> 00:19:34,640 But implementation is the true test. 265 00:19:35,060 --> 00:19:36,170 It's like a math exam. 266 00:19:36,170 --> 00:19:41,060 Rather than writing an essay, when you write an essay, you might be able to fool the teacher into 267 00:19:41,060 --> 00:19:42,460 thinking you understand something. 268 00:19:42,830 --> 00:19:46,810 But in a math exam, either your answers are right or your answers are wrong. 269 00:19:47,330 --> 00:19:51,260 In this context, either your code works or doesn't, it's very clear cut. 270 00:19:55,910 --> 00:20:01,940 When it comes to practicality, we cover that, too, we demonstrate our algorithms both on real world 271 00:20:01,940 --> 00:20:07,400 data like text and images, and we demonstrate our algorithms on two dimensional data sets so that you 272 00:20:07,400 --> 00:20:09,440 can visualize what an algorithm is doing. 273 00:20:10,160 --> 00:20:11,450 Both of these are important. 274 00:20:12,320 --> 00:20:15,490 Text and images are quite possibly the most practical data. 275 00:20:15,500 --> 00:20:20,690 There is some billion dollar companies exist and thrive because of those capabilities. 276 00:20:21,470 --> 00:20:26,170 Tuti data sets are critical as well because as humans, that's the only thing we can see. 277 00:20:26,720 --> 00:20:31,490 But at the same time, we don't try to pull the wool over your eyes and pretend like you need 20 different 278 00:20:31,490 --> 00:20:32,790 data sets to practice on. 279 00:20:33,260 --> 00:20:38,270 Instead, we learn something much more valuable how to use an algorithm on an infinite number of data 280 00:20:38,270 --> 00:20:42,620 sets, because we are able to abstract the idea that all data is the same. 281 00:20:43,250 --> 00:20:45,950 Remember that all the computer sees is a list of numbers. 282 00:20:45,950 --> 00:20:49,790 It doesn't know that those numbers represent height measurements or an image pixel. 283 00:20:50,660 --> 00:20:56,070 And of course, by experiencing both ends of the spectrum, both the theory and the implementation. 284 00:20:56,300 --> 00:21:01,370 And then finally plugging data into your model, that's going to make you very well-rounded professionally. 285 00:21:06,580 --> 00:21:11,910 To conclude this lecture, I want to pose a final question, how does all this knowledge help you? 286 00:21:12,640 --> 00:21:17,170 Why does understanding the approach of this course make you better at machine learning in general? 287 00:21:18,100 --> 00:21:22,360 Well, my hope is that this gives you a very high level bird's eye view of the landscape. 288 00:21:22,870 --> 00:21:26,170 You can see the different approaches of people from different backgrounds. 289 00:21:26,860 --> 00:21:30,850 Academic people typically don't see what other non-academic people are doing. 290 00:21:31,120 --> 00:21:35,500 And those people who hate math typically don't see what people who love math are doing. 291 00:21:35,920 --> 00:21:40,840 So with this bird's eye view, you can see everybody if you're learning machine learning, it helps 292 00:21:40,840 --> 00:21:46,540 you understand who am I competing with when I apply for a job and what kinds of jobs are out there and 293 00:21:46,540 --> 00:21:48,190 what would be adequate preparation. 294 00:21:48,730 --> 00:21:53,320 If you're hiring for a machine learning team, it helps you understand the different backgrounds of 295 00:21:53,320 --> 00:21:54,100 individuals. 296 00:21:54,100 --> 00:21:58,630 You have to compare to one another to understand their strengths and weaknesses and what they bring 297 00:21:58,630 --> 00:21:59,220 to the table.