1
00:00:02,960 --> 00:00:10,130
Hello, everyone, in this session, we are going to learn about machine learning concepts, OK, but

2
00:00:10,130 --> 00:00:15,160
before seeing the machine learning concepts, let's understand what is A.I.?

3
00:00:16,110 --> 00:00:24,080
Yeah, if you really see the computer software that thinks and acts like human beings, it it's a software,

4
00:00:24,090 --> 00:00:26,820
please know they can decide inside the hardware entity.

5
00:00:26,880 --> 00:00:29,610
OK, but it's essentially your software.

6
00:00:30,110 --> 00:00:31,490
It's a rapidly growing field.

7
00:00:31,500 --> 00:00:32,310
We all know that.

8
00:00:32,460 --> 00:00:32,870
Right.

9
00:00:32,900 --> 00:00:34,340
And the impact is already there.

10
00:00:34,890 --> 00:00:41,340
Some of the examples that you can think of, it is already being used to predict diseases, diseases

11
00:00:41,340 --> 00:00:46,220
like heart disease, cancer diseases, proactive detection.

12
00:00:46,290 --> 00:00:47,490
That's what we're talking about.

13
00:00:47,760 --> 00:00:52,200
It's being used to detect and predict diseases even before it occurs.

14
00:00:52,880 --> 00:00:55,550
It is used to detect fraud again proactively.

15
00:00:56,160 --> 00:00:59,720
It is used in recommendations as to what is this recommendation system.

16
00:01:00,420 --> 00:01:09,550
You watch a movie or listen to a song or watch a movie clip either on the Facebook or YouTube or any

17
00:01:09,550 --> 00:01:10,560
or voted up.

18
00:01:11,040 --> 00:01:17,620
After a while, you realize that the app is making some suggestions to you, right?

19
00:01:17,640 --> 00:01:18,690
You can watch this.

20
00:01:18,690 --> 00:01:20,100
You will enjoy this song.

21
00:01:20,730 --> 00:01:21,070
Right.

22
00:01:21,300 --> 00:01:22,230
So what is it?

23
00:01:23,880 --> 00:01:25,520
What is the suggestion based on?

24
00:01:25,590 --> 00:01:32,100
It is built on what is known as a recommendation system and recommendation system is built using artificial

25
00:01:32,100 --> 00:01:32,800
intelligence.

26
00:01:32,950 --> 00:01:33,330
Right.

27
00:01:33,810 --> 00:01:34,740
Face detection.

28
00:01:35,160 --> 00:01:43,470
Once upon a time, we used to mark attendance by signing and then we had access cards.

29
00:01:43,470 --> 00:01:50,850
Then the fingerprint reader came and now we have face addiction serving as a means to mark attendance

30
00:01:50,850 --> 00:01:54,120
and also to prevent intrusion.

31
00:01:54,120 --> 00:01:56,310
You know, for surveillance activities.

32
00:01:57,210 --> 00:02:00,350
We are using face detection to open apps.

33
00:02:00,360 --> 00:02:00,750
Right.

34
00:02:01,110 --> 00:02:03,770
So all this are applications of here.

35
00:02:03,780 --> 00:02:11,100
So that's why I said is already here and its impact is only going to increase in the months and years

36
00:02:11,100 --> 00:02:11,500
to come.

37
00:02:12,180 --> 00:02:12,570
Right.

38
00:02:12,990 --> 00:02:16,470
So having seen what is here is a software.

39
00:02:16,740 --> 00:02:17,070
Right.

40
00:02:17,370 --> 00:02:20,170
Let's understand the two different types of here.

41
00:02:20,730 --> 00:02:28,830
One is called strong A.I. and the other is called Vecchi, strong A.I. We are talking of systems that

42
00:02:28,830 --> 00:02:31,640
are thinking at the level of human beings.

43
00:02:32,230 --> 00:02:36,340
OK, that is that can think and act like human beings.

44
00:02:36,360 --> 00:02:36,700
Right.

45
00:02:37,860 --> 00:02:39,000
What you see in movies.

46
00:02:39,000 --> 00:02:39,260
Right.

47
00:02:39,270 --> 00:02:41,670
The fictional characters that you see in movies.

48
00:02:41,670 --> 00:02:41,990
Right.

49
00:02:42,000 --> 00:02:48,060
They are all talking of strong A.I. It is also called this artificial general intelligence and artificial

50
00:02:48,060 --> 00:02:49,080
superintelligence.

51
00:02:49,440 --> 00:02:53,160
But if you really see we are not there yet, OK?

52
00:02:53,370 --> 00:02:56,650
It's in the realm of research establishments, OK?

53
00:02:56,970 --> 00:03:00,420
People are scientists are actually researching this.

54
00:03:00,780 --> 00:03:07,740
We are far, far away from a world where all systems, artificial intelligence systems can think and

55
00:03:07,740 --> 00:03:09,090
act like human beings.

56
00:03:09,390 --> 00:03:11,100
OK, we are not there yet.

57
00:03:11,760 --> 00:03:13,500
So where are we today?

58
00:03:13,510 --> 00:03:19,440
We are in the realm of what is known as Vecchia, but don't get fooled by the term vecchia.

59
00:03:19,440 --> 00:03:27,240
It is only to make a distinction between strong A.I., right, the soup of the artificial super intelligence

60
00:03:27,240 --> 00:03:33,660
that we are calling the current level of artificial intelligence as we get into even with the so-called

61
00:03:33,660 --> 00:03:34,190
Vijaya.

62
00:03:34,200 --> 00:03:35,910
And we are solving a lot of problems.

63
00:03:35,910 --> 00:03:36,290
Right.

64
00:03:36,510 --> 00:03:37,830
But how are we doing it?

65
00:03:38,160 --> 00:03:41,340
We are solving problems by detecting patterns.

66
00:03:41,770 --> 00:03:43,980
OK, we are going to see what is a pattern.

67
00:03:43,980 --> 00:03:48,110
And please know that this is the dominant mode of A.I. today.

68
00:03:48,900 --> 00:03:53,750
OK, so which brings us to the question, what is the pattern?

69
00:03:54,980 --> 00:04:01,720
Right, a pattern is something that gets repeated in a recognizable way, OK?

70
00:04:01,920 --> 00:04:08,900
In fact, I will go a step further and say that pattern is a consistent, recurring characteristic.

71
00:04:10,010 --> 00:04:13,570
It's a consistent but a recurring characteristic.

72
00:04:13,880 --> 00:04:14,240
Right.

73
00:04:14,690 --> 00:04:21,050
So the system, the computer programming will identify something that gets repeated.

74
00:04:21,260 --> 00:04:21,650
Right.

75
00:04:21,650 --> 00:04:24,800
And uncover the pattern, you know, in your data.

76
00:04:24,800 --> 00:04:26,600
And through that, it will solve the problem.

77
00:04:26,990 --> 00:04:27,300
Right.

78
00:04:27,350 --> 00:04:29,280
That's how the world of A.I. is working today.

79
00:04:29,630 --> 00:04:33,260
And as I said earlier, this is the dominant mode of A.I..

80
00:04:34,120 --> 00:04:41,480
OK, so it is through these Vecchi systems, the so-called V.K. systems, that we are solving many,

81
00:04:41,480 --> 00:04:42,400
many problems.

82
00:04:43,010 --> 00:04:43,330
Right.

83
00:04:43,760 --> 00:04:46,220
So that's why I said Wikia is a misnomer.

84
00:04:46,370 --> 00:04:47,740
OK, don't get fooled by that.

85
00:04:48,020 --> 00:04:54,740
It's only in comparison to strong A.I. that we are categorizing the pattern recognition type of artificial

86
00:04:54,740 --> 00:04:59,220
intelligence, as Rekia is only for comparing and contrasting.

87
00:04:59,570 --> 00:04:59,870
Right.

88
00:05:00,410 --> 00:05:08,260
So now when you see this graph right, what are the things you can not just take a moment if you want,

89
00:05:08,270 --> 00:05:15,560
you can even fast for a while, right past the video to see what are the patterns you can note in this

90
00:05:15,560 --> 00:05:16,070
graph.

91
00:05:17,200 --> 00:05:26,170
Right, if you see the peaks are occurring in the month of July, which drops the lowest points are

92
00:05:26,170 --> 00:05:34,370
occurring in the month of January, the other pattern is the highest point in a calendar year.

93
00:05:34,610 --> 00:05:35,050
Right.

94
00:05:36,470 --> 00:05:38,660
Is progressively increasing.

95
00:05:39,500 --> 00:05:45,230
Right after nearly the same break, you see the difference between this and this, this and this, you

96
00:05:45,240 --> 00:05:48,170
know, it seems to be nearly at the same rate.

97
00:05:49,150 --> 00:05:49,460
Right.

98
00:05:49,780 --> 00:05:53,250
And the pattern gets repeated every calendar year.

99
00:05:54,990 --> 00:05:55,360
Right.

100
00:05:55,920 --> 00:06:02,990
So those are some of the patterns that we are able to uncover and we are able to uncover even with our

101
00:06:03,000 --> 00:06:03,270
audience.

102
00:06:03,270 --> 00:06:03,560
Right.

103
00:06:04,870 --> 00:06:09,620
But what if the number of data points are more likely, if there are millions of data points?

104
00:06:10,000 --> 00:06:11,330
How will you uncover the pattern?

105
00:06:12,010 --> 00:06:14,320
That is why you need the software, right?

106
00:06:14,350 --> 00:06:19,690
That is what an assistant will do, because in this, you are able to uncover the pattern even with

107
00:06:19,690 --> 00:06:20,290
your eyes.

108
00:06:21,340 --> 00:06:24,010
And you will see that for the next year.

109
00:06:24,580 --> 00:06:28,870
The highest point will be in July 16 and the lowest point will be January 16.

110
00:06:29,320 --> 00:06:32,700
And the highest point will be somewhere close to two hundred thousand.

111
00:06:33,430 --> 00:06:36,430
You can make such a rough estimates, right?

112
00:06:36,460 --> 00:06:38,530
You don't need even a calculator for that.

113
00:06:39,190 --> 00:06:42,430
You can do that even with simple manual calculations.

114
00:06:42,610 --> 00:06:42,930
Right.

115
00:06:43,300 --> 00:06:49,180
So we are able to uncover a pattern, you know, when through our eyes that just by observing if the

116
00:06:49,180 --> 00:06:53,980
number of data points or fewer, if the number of data points are more, we need systems.

117
00:06:53,980 --> 00:06:54,240
Right.

118
00:06:54,260 --> 00:06:56,970
That's what we are going to see, you know, in the course of this program.

119
00:06:59,440 --> 00:07:04,930
Now, let's see how is applied and this is prediction, right?

120
00:07:04,960 --> 00:07:06,520
I mentioned this example earlier.

121
00:07:08,320 --> 00:07:18,850
The typical database that is used to build the pattern will consist of patient records with sugar,

122
00:07:19,000 --> 00:07:27,250
B.P., heart disease, cholesterol, genetic predisposition, whether anyone in the family has had a

123
00:07:27,370 --> 00:07:28,510
heart disease or not.

124
00:07:29,160 --> 00:07:32,230
OK, BMI, body mass index.

125
00:07:32,620 --> 00:07:38,920
So these details of people who have had heart attack and people who have not had a heart attack will

126
00:07:38,920 --> 00:07:40,500
be there in our dataset.

127
00:07:40,510 --> 00:07:46,090
So the system will uncover the pattern contributing to heart disease.

128
00:07:46,090 --> 00:07:51,580
Probably people who have a heart disease have a certain level of a certain level of sugar, certain

129
00:07:52,020 --> 00:07:54,300
level of cholesterol, so on and so forth.

130
00:07:54,730 --> 00:08:00,370
So it is those values, right in combination is what is called a Zapater.

131
00:08:01,030 --> 00:08:09,730
So the system will uncover the pattern that is contributing to heart disease and venera in future.

132
00:08:10,270 --> 00:08:17,170
Any such values of BP, sugar, cholesterol, BMI, genetic predisposition and such values are fed into

133
00:08:17,170 --> 00:08:17,830
the system.

134
00:08:18,250 --> 00:08:24,580
The system will see if the pattern that it previously identified as contributing to the heart disease

135
00:08:24,910 --> 00:08:27,320
is present in the current dataset.

136
00:08:27,340 --> 00:08:32,440
If it is present, the system will see this individual is likely to develop a heart disease.

137
00:08:32,950 --> 00:08:37,130
If it is not there, the system will say the individual is not likely to develop heart disease.

138
00:08:37,540 --> 00:08:39,600
Please note that I'm using the word lightly.

139
00:08:40,080 --> 00:08:47,920
That is because we are talking of a probability think we are talking of a chance occurrence because

140
00:08:47,920 --> 00:08:53,890
we can't be 100 percent right in prediction, because we are not God and we are still human beings who

141
00:08:53,890 --> 00:08:55,680
have developed the system right.

142
00:08:55,690 --> 00:08:57,970
We are trying to play God, but we are not God.

143
00:08:57,970 --> 00:08:58,330
Right.

144
00:08:59,680 --> 00:09:05,950
But so are accuracy is never 100 percent so which is why we use the terminology likely to develop.

145
00:09:06,850 --> 00:09:13,720
So the same concept is what is applied in loan default or prediction also by someone who is consistent

146
00:09:13,720 --> 00:09:20,710
in paying the loan, you know, will have a regular stream of income, probably supported by his significant

147
00:09:20,710 --> 00:09:21,820
other income also.

148
00:09:22,180 --> 00:09:22,660
Right.

149
00:09:23,290 --> 00:09:27,820
As a stable job isn't a thriving career and so on and so forth.

150
00:09:27,850 --> 00:09:28,260
Right.

151
00:09:28,630 --> 00:09:32,830
Someone who is likely to default will exhibit the exact opposite.

152
00:09:32,840 --> 00:09:37,720
I probably experience the layoff rate works in an industry which is in a downfall.

153
00:09:37,730 --> 00:09:38,200
Right.

154
00:09:38,200 --> 00:09:39,190
And so on and so forth.

155
00:09:39,430 --> 00:09:44,190
So the algorithm, the computer program will try to uncover the pattern.

156
00:09:44,200 --> 00:09:44,430
Right.

157
00:09:44,470 --> 00:09:47,520
That is contributing to a loan default.

158
00:09:47,950 --> 00:09:54,130
Once the pattern is identified and I say pattern, it will it will create another set of values that

159
00:09:54,280 --> 00:09:58,990
taken together leads to a scenario of a loan default rate.

160
00:09:59,320 --> 00:10:06,600
So once that pattern is detected, it will check whether the same set of values are the same characteristic

161
00:10:06,610 --> 00:10:08,440
is there in the current dataset.

162
00:10:08,440 --> 00:10:14,140
Also, once it is there, it will say that this individual is likely to default.

163
00:10:14,140 --> 00:10:14,560
Right.

164
00:10:14,590 --> 00:10:16,940
So this is how the world of ER works.

165
00:10:17,380 --> 00:10:19,540
It tries to uncover the pattern.

166
00:10:20,610 --> 00:10:26,370
OK, from the given dataset, it tries to uncover the pattern and applies it for future.

167
00:10:26,760 --> 00:10:30,540
I hope you understand now how the world of works.

168
00:10:32,870 --> 00:10:33,260
Right.

169
00:10:33,530 --> 00:10:40,910
So having seen examples, having seen how it works, let's see some more aspects related to it.

170
00:10:41,690 --> 00:10:42,710
What do you see?

171
00:10:42,710 --> 00:10:47,620
You know, in in headlines and in media, it's all about the right.

172
00:10:47,960 --> 00:10:53,190
But if you see in real in this case itself, we are trying to learn about machine learning.

173
00:10:53,510 --> 00:10:55,430
So what is and what is machine learning?

174
00:10:55,880 --> 00:10:56,690
If you really see.

175
00:10:56,990 --> 00:10:57,380
Yeah.

176
00:10:57,380 --> 00:11:00,970
Is the headline that that's what is flashed in the media.

177
00:11:01,340 --> 00:11:04,630
But the technology that is driving A.I. is actually machine learning.

178
00:11:05,030 --> 00:11:05,410
Right.

179
00:11:05,870 --> 00:11:11,510
So all the pattern recognition that I talk about, the core underlying sciences, machine learning,

180
00:11:11,840 --> 00:11:17,690
and when whenever we talk of machine learning and artificial intelligence, you will also hear about

181
00:11:17,690 --> 00:11:20,450
another technology called deep learning.

182
00:11:20,450 --> 00:11:25,760
Deep learning, you know, use an advanced form of machine learning can be thought of as a subset of

183
00:11:26,600 --> 00:11:27,240
machine learning.

184
00:11:27,740 --> 00:11:27,980
Right.

185
00:11:28,250 --> 00:11:29,950
I hope you understand this.

186
00:11:31,930 --> 00:11:36,680
Now, let's understand machine learning in more detail using an example.

187
00:11:37,070 --> 00:11:46,840
OK, so I'm inputting three apples, the red colored apples, the green leaf, OK, it's in this particular

188
00:11:46,840 --> 00:11:52,180
shape, I'm giving three images and I'm referring to these three images as apples.

189
00:11:52,450 --> 00:11:57,910
I'm helping the computer system to understand that an object, if it looks like this, it is known as

190
00:11:57,910 --> 00:11:58,270
an apple.

191
00:11:58,870 --> 00:12:07,090
So the system, the model has been developed in such a way that if any object is input, it will check

192
00:12:07,090 --> 00:12:13,230
whether it these characteristics are displayed or the characteristics are displayed.

193
00:12:13,660 --> 00:12:14,830
It will say it is an apple.

194
00:12:16,270 --> 00:12:19,000
The characteristics are not they are not displayed.

195
00:12:20,140 --> 00:12:21,010
What will it say?

196
00:12:22,760 --> 00:12:23,780
Can you think of that?

197
00:12:25,490 --> 00:12:29,030
It will say it is not an apple, OK?

198
00:12:29,600 --> 00:12:31,550
Why does it say it's not an apple?

199
00:12:32,750 --> 00:12:40,940
Because all that the system knows is if these characteristics are there, the red color, little green

200
00:12:40,940 --> 00:12:47,090
leaf, right, this particular shape, if these characteristics are there, it is known as an apple

201
00:12:47,750 --> 00:12:51,350
in these characteristics are not there is not an apple.

202
00:12:52,280 --> 00:12:52,470
Right.

203
00:12:53,210 --> 00:12:55,970
What if I feed this particular apple?

204
00:12:57,740 --> 00:13:00,890
If I feed this particular apple, what will the system say?

205
00:13:02,920 --> 00:13:11,700
It will still say it is not an Apple one, because in this the color is different, right?

206
00:13:11,740 --> 00:13:18,760
This is pink, this pinkish apple Pinkel an apple that has this is a dark red apple, right, 100.

207
00:13:20,000 --> 00:13:26,330
OK, now let's see one more example, if I feed this image of the apple, what will the system say?

208
00:13:28,970 --> 00:13:31,730
Will it say it's an apple or not an Apple thing?

209
00:13:33,320 --> 00:13:35,720
It will say it is not an apple.

210
00:13:39,710 --> 00:13:47,750
Because the leaf, right, the the proportion of leaf is on the right, but here, if you see the progression

211
00:13:47,750 --> 00:13:48,920
of the leaf is on the left.

212
00:13:50,950 --> 00:13:52,200
We are human beings, right?

213
00:13:53,020 --> 00:13:57,590
We know that the pollution can be on the right or left on both sides.

214
00:13:58,120 --> 00:14:01,550
The system doesn't know, the system knows based on what you're fed.

215
00:14:01,720 --> 00:14:08,740
What we are fed is a dark colored apple with this particular shape and with the conclusion of leaf on

216
00:14:08,740 --> 00:14:09,270
the left side.

217
00:14:10,400 --> 00:14:14,270
Those are the characteristics that we're fed and we have labeled them as apples.

218
00:14:15,200 --> 00:14:18,030
But here, the leaf is protruding on the right.

219
00:14:18,140 --> 00:14:26,690
So the system will say that it is not an apple because the characteristic of protrusion of leaf is on

220
00:14:26,690 --> 00:14:28,700
the right and not on the left right.

221
00:14:29,660 --> 00:14:30,980
What if I feed an orange?

222
00:14:31,430 --> 00:14:32,900
And you said it's not an apple?

223
00:14:33,920 --> 00:14:34,260
Right.

224
00:14:34,460 --> 00:14:40,640
So this is known as what is called a supervised learning, that is a reference is provided.

225
00:14:41,920 --> 00:14:48,520
If an object were to look like this, I'm giving the reference, OK, or I'm giving a label known as

226
00:14:48,520 --> 00:14:49,330
apples.

227
00:14:50,040 --> 00:14:54,640
OK, if those references are annotations or labels are not there.

228
00:14:56,800 --> 00:15:05,080
The system will try to group similar type of objects based on the characteristics you see, I have fed

229
00:15:05,770 --> 00:15:10,700
images of apples, the brightly colored, the pink colored and the bananas.

230
00:15:10,720 --> 00:15:11,060
Right.

231
00:15:11,470 --> 00:15:15,730
So the system has grouped the red coloured apples together.

232
00:15:15,730 --> 00:15:20,090
The pink coloured apples are grouped together and bananas the group together.

233
00:15:20,770 --> 00:15:27,900
So this type of machine learning is known as unsupervised learning because the references or labels

234
00:15:27,910 --> 00:15:28,470
are not.

235
00:15:29,660 --> 00:15:37,280
If you see the the the models that are used today, majority of the models are in the realm of supervised

236
00:15:37,280 --> 00:15:38,030
learning only.

237
00:15:38,580 --> 00:15:45,470
OK, the example that we saw this is prediction and loan default prediction would come under the area

238
00:15:45,470 --> 00:15:47,330
of supervised learning only.

239
00:15:47,690 --> 00:15:50,480
OK, then why do we have unsupervised learning?

240
00:15:51,080 --> 00:15:55,160
Unsupervised learning is used in the area of segmenting customers.

241
00:15:57,170 --> 00:16:00,710
Normal way of segmenting customers is based on one or two dimensions.

242
00:16:00,980 --> 00:16:06,230
Based on the revenue the customer generates are based on the type of the customer.

243
00:16:06,230 --> 00:16:12,160
The customer is a large customer of small and medium business customer or an individual retail customer.

244
00:16:12,410 --> 00:16:12,770
Right.

245
00:16:12,830 --> 00:16:14,120
So those are small things.

246
00:16:14,120 --> 00:16:20,860
But what a systems can do is uncover new segments and that wealth of information for any organization.

247
00:16:21,230 --> 00:16:21,470
Right.

248
00:16:21,860 --> 00:16:28,420
So supervisor unsupervised learning or what are used extensively as as of today.

249
00:16:28,430 --> 00:16:28,730
Right.

250
00:16:29,030 --> 00:16:35,180
And even between the two supervised learning is what is used more than unsupervised learning because

251
00:16:35,180 --> 00:16:39,140
unsupervised learning is primarily used when references are not there.

252
00:16:39,890 --> 00:16:47,810
The third type of reinforcement, the third type of machine learning that is also applied but not as

253
00:16:48,140 --> 00:16:51,400
widely used as supervised learning is reinforcement learning.

254
00:16:51,440 --> 00:16:55,790
You're the system learns from mistakes and responses.

255
00:16:56,030 --> 00:17:00,200
If you feed an apple, it will probably say a mango, right?

256
00:17:00,200 --> 00:17:02,960
Then you say, no, it's not a mango, it's an apple.

257
00:17:03,050 --> 00:17:03,360
Right.

258
00:17:03,860 --> 00:17:05,630
System notes it right.

259
00:17:05,840 --> 00:17:12,970
And this kind of cycle goes on and on, probably tens and thousands of cycles.

260
00:17:12,980 --> 00:17:17,840
OK, the system learns and then it correctly says that it is an apple.

261
00:17:18,470 --> 00:17:18,870
Right.

262
00:17:19,040 --> 00:17:23,030
So this is known as reinforcement learning, as I said, is not as widely used.

263
00:17:23,520 --> 00:17:28,400
OK, so the one that is widely used is supervised learning, unsupervised learning.

264
00:17:28,400 --> 00:17:32,460
As I said, it's used primarily in the world of customer segmentation.

265
00:17:33,200 --> 00:17:35,660
What is the other benefit of unsupervised learning?

266
00:17:37,530 --> 00:17:45,850
It is also used to detect outliners fraud detection, unsupervised learning is used right.

267
00:17:46,140 --> 00:17:51,660
The biggest advantage with unsupervised learning is you don't need these references or annotations or

268
00:17:51,670 --> 00:17:52,170
labels.

269
00:17:52,410 --> 00:17:52,760
Right.

270
00:17:52,920 --> 00:17:57,600
So this is something that we saw earlier, machine learning, the three types of machine learning,

271
00:17:57,600 --> 00:18:00,510
supervised and supervised and reinforced.

272
00:18:01,780 --> 00:18:09,100
Right, there's one more way in which you can classify or categorize machine learning, what is called

273
00:18:09,460 --> 00:18:15,280
classification type of machine learning or regression type of machine, but we also call it a classification

274
00:18:15,280 --> 00:18:17,170
algorithm or regression algorithm.

275
00:18:17,650 --> 00:18:19,420
Let's understand this with an example.

276
00:18:20,550 --> 00:18:26,340
Right, in the case of regression, we are telling what is the temperature going to be?

277
00:18:27,360 --> 00:18:27,670
Right.

278
00:18:28,030 --> 00:18:34,950
It's going to be eighty four degree Fahrenheit if I see whether it'll be hot or cold based on the temperature,

279
00:18:35,650 --> 00:18:36,060
right.

280
00:18:37,080 --> 00:18:42,780
Based on the temperature for tomorrow, then I am talking of classification, if you are to extrapolate

281
00:18:42,780 --> 00:18:50,100
this to the student example, if I tell how many months the student is going to get OK, then I am talking

282
00:18:50,100 --> 00:18:50,970
of regression.

283
00:18:51,510 --> 00:18:57,330
If I say whether the student will score high marks or low marks or medium range marks, then I am talking

284
00:18:57,330 --> 00:18:58,590
of classification.