1 00:00:00,920 --> 00:00:08,090 Let's go back to this topic of machines and computers and the fact that they don't really understand 2 00:00:08,180 --> 00:00:09,700 words right. 3 00:00:09,710 --> 00:00:13,850 If I ask my computer hey is this a snake. 4 00:00:13,850 --> 00:00:15,770 Is this a python. 5 00:00:15,800 --> 00:00:22,920 My computer is going to well I have no idea what I'm talking about when I look at an image of a snake 6 00:00:23,790 --> 00:00:31,380 no matter what color it is no matter whether it's a cartoon or real life image or whether it has different 7 00:00:31,380 --> 00:00:33,750 directions with different animals in it. 8 00:00:33,810 --> 00:00:45,190 I can identify a snake right away our brains just know but to tell my computer what a snake is is almost 9 00:00:45,280 --> 00:00:46,940 impossible right. 10 00:00:47,020 --> 00:00:49,610 Because there's so many factors. 11 00:00:49,720 --> 00:00:59,200 A snake doesn't have legs and arms but an airplane doesn't have legs and arms a snake has scales. 12 00:00:59,200 --> 00:01:06,480 But then the computer is going to ask me what are scales but with machine learning we can actually teach 13 00:01:06,840 --> 00:01:14,670 a computer to know what a snake is or at least pretended like it knows. 14 00:01:14,670 --> 00:01:17,190 So how do we do this. 15 00:01:17,200 --> 00:01:25,880 Well you remember this diagram I hope it's simply what we do in programming that is we give an input 16 00:01:25,970 --> 00:01:34,430 let's say a list of one two three we give it to a function and there's an output of 2 4 6. 17 00:01:34,430 --> 00:01:44,480 So clearly this function multiplies our list items in our list by 2 and often our programs are procedural 18 00:01:45,560 --> 00:01:46,910 that is we have. 19 00:01:46,910 --> 00:01:53,780 If this then that do this then this type of conditions we tell the computer what to do. 20 00:01:54,050 --> 00:02:03,910 We write as programmers the input and the function and all the computer does is actually just calculate 21 00:02:03,970 --> 00:02:06,880 using the input and the function the output. 22 00:02:06,880 --> 00:02:13,300 And it can do that really really fast for us instead of having a human do something when let's say a 23 00:02:13,300 --> 00:02:18,570 list is millions of items long machine learning changes. 24 00:02:18,570 --> 00:02:30,730 This idea it says hey give me the input and give me what you want the output to be that is give me all 25 00:02:30,730 --> 00:02:32,500 the data that you have. 26 00:02:32,500 --> 00:02:35,410 And by the way tell me what the answer is. 27 00:02:35,470 --> 00:02:37,870 Or at least what the output that you want. 28 00:02:37,870 --> 00:02:46,090 So in this case I want the input to be a list of 1 2 and 3 and the output to be a list of 2 4 and 6 29 00:02:47,800 --> 00:02:57,460 and then the interesting part is that as developers we give this input and output but with machine learning 30 00:02:57,520 --> 00:03:02,630 we say hey machine you figure out what this function is. 31 00:03:03,070 --> 00:03:12,790 Because if I give you let's say instead of a list A list of all these snake pictures or maybe animals 32 00:03:13,000 --> 00:03:17,020 and I want you to tell me if this image has a snake or not. 33 00:03:17,020 --> 00:03:21,210 Or maybe these 100 images which one of them have snakes. 34 00:03:21,250 --> 00:03:25,010 Well I don't know how to tell you how to do that. 35 00:03:25,030 --> 00:03:28,590 So instead you figure it out yourself. 36 00:03:28,720 --> 00:03:31,000 Here are all the pictures. 37 00:03:31,000 --> 00:03:39,850 And here is what I expect the output to be either snake or not snake and with machine learning the function 38 00:03:40,000 --> 00:03:44,280 is generated or created by the machine. 39 00:03:44,350 --> 00:03:46,830 Sounds really cool doesn't it. 40 00:03:46,870 --> 00:03:53,040 And this is machine learning it's simply a function. 41 00:03:53,150 --> 00:03:59,810 So instead of us writing the function like we have before with machine learning the machine writes the 42 00:03:59,810 --> 00:04:06,770 function we just give it the input and the output and the word function can be replaced with a model. 43 00:04:06,890 --> 00:04:09,770 You might have heard of the word machine learning model. 44 00:04:09,770 --> 00:04:11,320 That's all a model is. 45 00:04:11,450 --> 00:04:19,740 It's a function that receives an input and returns an output another word might be algorithm a machine 46 00:04:19,830 --> 00:04:26,370 learning algorithm a machine learning bought and at the end of the day it's a brain we're pretending 47 00:04:26,550 --> 00:04:34,330 that this machine is a brain that understands the inputs and gives us the desired output. 48 00:04:34,330 --> 00:04:41,080 Now the interesting side effect of this is that often when we talk about machine learning models these 49 00:04:41,170 --> 00:04:48,610 functions are really hard to understand because the computer comes up with this function as a human. 50 00:04:48,700 --> 00:04:51,710 It's hard to really understand how it works. 51 00:04:51,820 --> 00:04:58,720 And as these models get more and more complex it becomes harder and harder for us as developers to understand 52 00:04:58,750 --> 00:04:59,790 how it works. 53 00:05:00,010 --> 00:05:06,430 Which raises a lot of issues but we don't need to get into that now as you can see the main part of 54 00:05:06,430 --> 00:05:15,310 this is that in order to train this machine to create this function we have to give it data and for 55 00:05:15,310 --> 00:05:16,600 it to be really really good. 56 00:05:16,690 --> 00:05:24,060 We have to give it lots and lots of data and then we want to test to make sure that this data that we 57 00:05:24,060 --> 00:05:26,530 give in gives us the correct output. 58 00:05:27,370 --> 00:05:34,420 So at the end of the day machine learning is simply a computer writing its own function based on the 59 00:05:34,420 --> 00:05:36,550 inputs and outputs. 60 00:05:36,580 --> 00:05:38,280 That's pretty cool isn't it. 61 00:05:38,290 --> 00:05:39,900 I'll see you in the next video. 62 00:05:40,060 --> 00:05:40,300 Bye.