1 00:00:00,760 --> 00:00:01,170 Hello. 2 00:00:01,230 --> 00:00:02,070 Welcome back. 3 00:00:02,070 --> 00:00:05,430 So currently our image is still not a column vector. 4 00:00:05,430 --> 00:00:08,100 We saw the dimension of our image. 5 00:00:08,100 --> 00:00:08,710 Let's see. 6 00:00:10,230 --> 00:00:13,970 So this the dimension the dimension of our images still. 7 00:00:14,210 --> 00:00:17,180 Um well this the dimension of the entire set. 8 00:00:17,190 --> 00:00:21,750 But we know the dimension of the image at least is sixty four by sixty four by three. 9 00:00:21,750 --> 00:00:24,720 We've got to change this into a column vector. 10 00:00:24,720 --> 00:00:27,570 We've got to change this into just this. 11 00:00:27,570 --> 00:00:34,350 We have got to make it like this mean in one column and then over 12000 rows. 12 00:00:34,560 --> 00:00:41,280 And once we do this for a single image for the entire image sets we're going to have these number of 13 00:00:41,280 --> 00:00:46,230 columns number of columns will be equal to the number of training examples a number of rows will be 14 00:00:46,230 --> 00:00:47,280 equal to this number. 15 00:00:47,310 --> 00:00:50,580 We have to flatten each image to a column vector. 16 00:00:50,670 --> 00:00:51,690 That is what we have to do now. 17 00:00:51,720 --> 00:00:53,740 That's the pre processing we need to do. 18 00:00:53,760 --> 00:01:00,410 So let's do the math and I'm going to comment this. 19 00:01:00,430 --> 00:01:01,890 I'm going to say it again 20 00:01:04,950 --> 00:01:08,970 call this number three and then 21 00:01:11,890 --> 00:01:16,190 I'm going to form this once we are done we can clean the format we can clean the comment. 22 00:01:16,450 --> 00:01:23,320 Yeah but I think if it's your first time doing this will be nice to have the test point for test it 23 00:01:23,830 --> 00:01:28,660 before it becomes complicated all of a sudden yeah. 24 00:01:28,780 --> 00:01:29,140 So 25 00:01:32,100 --> 00:01:33,870 so we going to come down here 26 00:01:37,290 --> 00:01:39,390 and then I'm going to say 27 00:01:42,130 --> 00:01:54,360 Qur'an set X flatten we're going to flatten the image and all we're gonna do is take train sets X original 28 00:01:55,890 --> 00:02:02,370 and then we're going to use a function on us reshape so a dot reshape 29 00:02:05,040 --> 00:02:17,060 what we want to reshape is train set one or reshape it to this train set X original index 0 and then 30 00:02:17,190 --> 00:02:24,690 minus 1 and then the transpose of this right we're gonna do the same for test set X 31 00:02:41,860 --> 00:02:44,050 and go see the 32 00:02:48,750 --> 00:02:50,670 right index here over here 33 00:02:53,790 --> 00:03:00,160 and then transpose once that is done we can print out the result and see what we have 34 00:03:11,480 --> 00:03:22,380 you can see a new image shape is then we can just use the STAR function and then I'm simply going to 35 00:03:22,380 --> 00:03:22,950 call this 36 00:03:32,490 --> 00:03:34,280 the sentence does and so on repos 37 00:03:38,970 --> 00:03:44,410 Chris set X shape is this okay. 38 00:03:44,590 --> 00:03:46,330 And then we're going to 39 00:03:48,970 --> 00:03:49,990 check the tests at 40 00:03:59,860 --> 00:04:00,820 right. 41 00:04:00,830 --> 00:04:04,400 We already saw the um the shape of the of the Y. 42 00:04:04,400 --> 00:04:06,000 We know the shape of the Y already. 43 00:04:06,030 --> 00:04:14,690 It was one by two hundred and nine which is the um which is the number of images oh just this you'll 44 00:04:14,700 --> 00:04:20,720 see I'm going to copy this and bring it over here. 45 00:04:20,740 --> 00:04:22,140 Nice view this as well. 46 00:04:22,640 --> 00:04:30,180 If you forgot and then I'm going to copy this controversy to copy put it over here. 47 00:04:30,260 --> 00:04:35,600 This has to be screwed contra s when I run this run module 48 00:04:40,580 --> 00:04:44,320 only into just color arrays can be converted to escalate index. 49 00:04:44,330 --> 00:04:45,860 Okay let's see where the error lies. 50 00:04:49,710 --> 00:04:50,000 Yeah. 51 00:04:50,140 --> 00:04:57,450 So over here I forgot to change the word here to test this front and see what up this is the cause of 52 00:04:57,450 --> 00:05:00,300 the arrow run module. 53 00:05:01,470 --> 00:05:01,860 Okay. 54 00:05:01,860 --> 00:05:06,170 Green set X flatten train set X or reshape the train set. 55 00:05:06,190 --> 00:05:07,380 X okay 56 00:05:15,540 --> 00:05:15,930 yeah. 57 00:05:15,980 --> 00:05:17,690 The courses from here. 58 00:05:17,690 --> 00:05:24,390 I forgot I was extracting the shape that's index 0. 59 00:05:24,430 --> 00:05:24,750 Okay. 60 00:05:24,770 --> 00:05:30,350 Control s run module. 61 00:05:30,860 --> 00:05:35,580 Kind of reshape array of size this into shape of this okay. 62 00:05:38,180 --> 00:05:38,900 Um yeah. 63 00:05:38,940 --> 00:05:39,560 Sorry about that. 64 00:05:39,570 --> 00:05:40,580 It's my fault. 65 00:05:40,650 --> 00:05:45,730 During my copy and paste in this is the mistake I made over here. 66 00:05:45,780 --> 00:05:52,050 I should be typing I should create a new variable I meant to create a new variable by changing this 67 00:05:52,050 --> 00:05:53,970 word here to test. 68 00:05:54,180 --> 00:05:58,500 So it's as if we were trying to reshape Um this this one again. 69 00:05:58,500 --> 00:06:00,640 That is why we received the arrow. 70 00:06:00,870 --> 00:06:01,400 Let's see now. 71 00:06:01,410 --> 00:06:08,010 So we have train sets X flatten test set X flatten and we have you in train sets X flatten test sets 72 00:06:08,130 --> 00:06:09,250 X flatten. 73 00:06:09,420 --> 00:06:11,260 Okay. 74 00:06:11,580 --> 00:06:12,330 Run module 75 00:06:15,130 --> 00:06:16,680 we still get in the same area. 76 00:06:16,690 --> 00:06:20,950 Test set X flatten test set on a score. 77 00:06:20,980 --> 00:06:25,740 Original 3 shape. 78 00:06:27,920 --> 00:06:29,070 Uh yeah. 79 00:06:29,100 --> 00:06:30,630 There with me. 80 00:06:30,630 --> 00:06:32,280 It's from here. 81 00:06:32,300 --> 00:06:36,120 Um I'm sure some of you have a go at me in the review section. 82 00:06:36,120 --> 00:06:36,840 Just bear with me. 83 00:06:36,850 --> 00:06:37,110 Yeah. 84 00:06:37,330 --> 00:06:38,080 Um yeah. 85 00:06:39,050 --> 00:06:39,640 So yeah. 86 00:06:39,790 --> 00:06:40,080 Yeah. 87 00:06:40,080 --> 00:06:41,210 This the last type. 88 00:06:41,280 --> 00:06:43,450 Yeah. 89 00:06:43,950 --> 00:06:45,050 Okay let's see now. 90 00:06:45,660 --> 00:06:46,190 Okay. 91 00:06:46,200 --> 00:06:46,610 Voila. 92 00:06:46,650 --> 00:06:48,690 Yeah we go so we don't have a go at me. 93 00:06:48,690 --> 00:06:49,030 Yeah. 94 00:06:49,590 --> 00:06:49,940 Okay. 95 00:06:50,010 --> 00:06:51,790 So um let's see. 96 00:06:51,790 --> 00:06:58,170 Train set X shape is this and then it's printing out into shape. 97 00:06:58,170 --> 00:06:59,550 Okay we'll fix that later. 98 00:07:00,060 --> 00:07:06,570 So we have the Y shape we know this already the Y set is one by two hundred and nine one by number of 99 00:07:06,600 --> 00:07:08,330 images we saw this. 100 00:07:08,340 --> 00:07:11,210 Let's see if our image shows this. 101 00:07:11,220 --> 00:07:18,210 We said we're going to take Y we're going to take all of y values and store them in a vector like this. 102 00:07:18,240 --> 00:07:20,950 So it's one dimensional in a way. 103 00:07:22,140 --> 00:07:27,120 So one by two or nine we have this stored like this. 104 00:07:27,120 --> 00:07:33,600 This is stored in a real vector we have we have multiple rows single. 105 00:07:33,990 --> 00:07:40,510 This is stored in a row vector we have multiple columns a single row and this is the one for the test. 106 00:07:40,560 --> 00:07:41,650 Okay. 107 00:07:41,730 --> 00:07:42,440 Right. 108 00:07:42,540 --> 00:07:42,840 Okay. 109 00:07:42,870 --> 00:07:43,910 Let's shorten this 110 00:07:46,970 --> 00:07:53,120 okay so what I meant to print was Dot shape not the entire thing. 111 00:07:53,510 --> 00:07:53,910 Okay. 112 00:07:53,930 --> 00:07:55,130 Control is to save 113 00:08:00,150 --> 00:08:02,780 Okay so that's the shape of our train set. 114 00:08:02,850 --> 00:08:07,070 X this is the shape now by the number of images. 115 00:08:07,070 --> 00:08:09,120 This the dimension of a single image. 116 00:08:09,140 --> 00:08:12,670 So this here represents the number of rows we have. 117 00:08:12,680 --> 00:08:18,860 This represents the number of columns and we can verify that from our image. 118 00:08:18,860 --> 00:08:19,360 Right. 119 00:08:19,380 --> 00:08:27,730 As you can see a single column starts from zero and at twelve thousand two hundred and eighty eight. 120 00:08:28,550 --> 00:08:36,800 And we've got multiple columns the number of columns we call equal to m m means the number of training 121 00:08:36,800 --> 00:08:38,040 examples. 122 00:08:38,060 --> 00:08:42,710 And over here we have two hundred and nineteen examples hence we have twelve thousand two hundred and 123 00:08:42,740 --> 00:08:44,870 eighty eight by two hundred and nine. 124 00:08:44,880 --> 00:08:50,380 And we have the same image the same image type but this time 50 of them for testing. 125 00:08:50,840 --> 00:08:53,300 And then we spoke about the Y already. 126 00:08:53,360 --> 00:08:53,750 So yeah. 127 00:08:53,750 --> 00:08:59,420 This the pre processing required in order to to feed the image to our neural network. 128 00:08:59,420 --> 00:09:05,600 So these this the flattened version is going to be the the version we feed into the neural network. 129 00:09:05,600 --> 00:09:09,740 Like we discussed in the theoretical class this all there is for this lesson if you have any questions 130 00:09:09,740 --> 00:09:10,040 at all. 131 00:09:10,040 --> 00:09:10,810 Just let me know. 132 00:09:10,940 --> 00:09:16,580 And sorry for the um the typos and all the other stuff but certainly you can have a go at me in the 133 00:09:16,580 --> 00:09:19,440 review section and um simulator. 134 00:09:19,440 --> 00:09:20,380 Have a nice day.