File size: 4,159 Bytes
d157f08
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
1
00:00:00,530 --> 00:00:01,090
I guess.

2
00:00:01,110 --> 00:00:05,220
And welcome to chapter eight point five where we talk about what one encoding.

3
00:00:05,520 --> 00:00:11,900
So as we saw before we did some transmissions on the extreme and X deciliter that's so image data.

4
00:00:11,970 --> 00:00:15,520
Now what about our label little white green and white test.

5
00:00:15,780 --> 00:00:16,530
Well let's find out.

6
00:00:16,530 --> 00:00:19,790
So let's recap what we did on the image data.

7
00:00:19,860 --> 00:00:23,880
We added a four dimension to go from this to this.

8
00:00:23,880 --> 00:00:32,070
We changed it to flow to two data type and we normalized it between 0 and 1 by dividing by 255.

9
00:00:32,450 --> 00:00:33,960
But what do we do to label that.

10
00:00:33,960 --> 00:00:35,450
Now that's true.

11
00:00:35,940 --> 00:00:39,730
So the label is basically in this form for white train.

12
00:00:39,750 --> 00:00:47,850
It's going to be a matrix that is 60000 that has 60000 elements and each element indicates a class label.

13
00:00:47,890 --> 00:00:57,360
So for x for y treatment which is a 28 if this element in white and extreme is going to be 28 by 28

14
00:00:57,840 --> 00:01:00,900
and this zero corresponds to its label here.

15
00:01:01,150 --> 00:01:04,710
However Harris does not use label data like this.

16
00:01:04,890 --> 00:01:06,820
It needs it to be a hot one encoded.

17
00:01:06,900 --> 00:01:09,850
And what does that look like that looks like this.

18
00:01:10,080 --> 00:01:11,640
So we have labels here.

19
00:01:12,030 --> 00:01:20,840
And instead of having a for being represented here it's basically a matrix to has 10 columns now.

20
00:01:21,270 --> 00:01:24,900
So instead of having no call them so effectively want to call them.

21
00:01:24,930 --> 00:01:26,890
Sorry sixty dozen columns.

22
00:01:27,000 --> 00:01:34,580
It has 10 columns here and 60000 rows and each column is a row.

23
00:01:34,590 --> 00:01:39,390
Sorry has basically a 1 0 0 indicating which label it is.

24
00:01:39,690 --> 00:01:42,900
So imagine we have this being transformed.

25
00:01:42,900 --> 00:01:44,090
Sorry let's look at this.

26
00:01:44,090 --> 00:01:45,520
This is a td rule here.

27
00:01:45,810 --> 00:01:48,340
Being transformed into this.

28
00:01:48,370 --> 00:01:55,520
So instead of having this rule before what one coding makes it into this I hope you understand clearly

29
00:01:55,530 --> 00:01:56,600
so we're going to do this now.

30
00:01:56,640 --> 00:01:58,770
You know I write in my book.

31
00:01:59,650 --> 00:01:59,960
OK.

32
00:01:59,970 --> 00:02:06,310
So Step three is a hot one including a full y labels and to do this we basically use any utilities that

33
00:02:06,310 --> 00:02:10,280
are imported from the utilities and all this stuff here.

34
00:02:10,290 --> 00:02:12,600
It just sends it to categorical.

35
00:02:12,600 --> 00:02:17,210
That is how Cara's calls dysfunction Hotpoint including two categorical.

36
00:02:17,460 --> 00:02:18,550
So we have Whitopia.

37
00:02:18,620 --> 00:02:23,160
It's equal to utilities not to categorical and just put the wager in here.

38
00:02:23,310 --> 00:02:25,140
And that transforms it.

39
00:02:25,140 --> 00:02:27,220
So let's take a look at how this actually looks.

40
00:02:27,220 --> 00:02:28,510
So let's run this here.

41
00:02:28,820 --> 00:02:30,120
So no why train.

42
00:02:30,270 --> 00:02:32,420
Let's look at the first rule in waitron.

43
00:02:32,830 --> 00:02:40,700
It's this and basically can see one two three four five six seven eight nine ten elements.

44
00:02:40,920 --> 00:02:45,270
And with this one this looks like the nine and fifth element here.

45
00:02:45,300 --> 00:02:46,920
So this is number five.

46
00:02:46,920 --> 00:02:51,720
So the first element in overtreating data is number five.

47
00:02:52,110 --> 00:02:54,710
So now let's move on to creating a model.