1 00:00:00,390 --> 00:00:05,460 Now, let's see which areas can we use artificial neural networks. 2 00:00:05,490 --> 00:00:08,100 These are the techniques of neural networks. 3 00:00:08,430 --> 00:00:15,900 You have classification neural networks and neural nets for can be trained to classify given pattern 4 00:00:15,900 --> 00:00:19,310 or data set into predefined classes. 5 00:00:19,950 --> 00:00:23,130 It uses feet forward for network for this purpose. 6 00:00:24,270 --> 00:00:26,820 Another one that we have here is prediction. 7 00:00:26,820 --> 00:00:34,230 Neural network and neural network can be trained to predict and outputs that are expected from giving 8 00:00:34,230 --> 00:00:34,780 input. 9 00:00:35,220 --> 00:00:42,240 We have several examples here, like a stock market prediction, weather prediction, traffic prediction 10 00:00:42,330 --> 00:00:43,110 and more. 11 00:00:43,530 --> 00:00:51,930 Next one to use clustering neural network and vector quantization clustering neural network is just 12 00:00:51,930 --> 00:00:59,400 a network that can be used to identify a unique feature of the data and then classify them into different 13 00:00:59,400 --> 00:01:05,140 categories without any prior knowledge of the data in this area. 14 00:01:05,160 --> 00:01:13,330 We have different algorithm like competitive networks, adaptive reason and temporary networks covered 15 00:01:13,330 --> 00:01:15,550 in self organizing maps and more. 16 00:01:16,470 --> 00:01:23,610 But let me just give you a better example for clarifying their purposes. 17 00:01:24,360 --> 00:01:32,560 Suppose you have ten samples of tweeze or geographical places or 10 different patients. 18 00:01:32,970 --> 00:01:35,460 We have some information from them. 19 00:01:35,790 --> 00:01:43,160 And these systems, these methods, these techniques are helping us to classify them with a number. 20 00:01:43,590 --> 00:01:50,690 So classifying them into different sets and finally represent each feature of them with a number. 21 00:01:50,880 --> 00:01:52,410 That's what we call here. 22 00:01:52,410 --> 00:01:59,720 Victor Quantization, costarring neural network vector quantization and classifying neural networks. 23 00:01:59,730 --> 00:02:08,310 They are pretty the same, which of course some differences, but we can use all of them for classifications. 24 00:02:09,060 --> 00:02:10,800 Next one is feature extension. 25 00:02:10,800 --> 00:02:12,130 Any major coordination. 26 00:02:12,690 --> 00:02:20,550 Let's say you are driving in a highway and there are some cameras to detect the plate number of your 27 00:02:20,550 --> 00:02:20,970 car. 28 00:02:24,180 --> 00:02:36,090 This is an example of a plate number, for example, here we have a B and then seven to zero, Avago 29 00:02:36,090 --> 00:02:45,660 is giving somehow consciousness to this camera to detect, to understand this plate number, to take 30 00:02:45,660 --> 00:02:55,350 the last digit and decide to issue a penalty for this car or not based on these last number and decoding 31 00:02:55,350 --> 00:03:04,720 data, if the last digit of the plate number is zero and one, then this car cannot drive on Monday. 32 00:03:05,430 --> 00:03:12,160 The first thing for this camera is to detect is zero, because this is just an image for a camera. 33 00:03:12,160 --> 00:03:15,600 A camera doesn't have any understanding of this image. 34 00:03:16,720 --> 00:03:23,860 We can train the camera using different image recognition techniques to detect that this is a zero. 35 00:03:24,330 --> 00:03:32,100 After that, the camera needs to decide that this car can go outside based on this plate number or not. 36 00:03:32,430 --> 00:03:35,340 This is the feature extraction part. 37 00:03:35,470 --> 00:03:40,370 So we are just extracting a feature from this model. 38 00:03:41,310 --> 00:03:44,430 Next one that we have here is association. 39 00:03:44,440 --> 00:03:45,660 Neural networks. 40 00:03:46,830 --> 00:03:55,170 Neural network can be trained to remember the particular pattern so that when the noise pattern's presented 41 00:03:55,440 --> 00:04:05,280 to the network, the network associates it with the closest one in the memory or disregarded Hockfield 42 00:04:05,280 --> 00:04:10,390 networks which perform recognition, classification and clustering and more. 43 00:04:11,220 --> 00:04:18,330 We can use these techniques for reading different handwritings, for example, and just writing LATAPY 44 00:04:18,360 --> 00:04:18,990 like that. 45 00:04:20,140 --> 00:04:28,320 There are several sets in these networks memory and the network is going to compare it with one of the 46 00:04:28,320 --> 00:04:30,540 models on its memory. 47 00:04:31,350 --> 00:04:34,780 But what if I won't let her be like that? 48 00:04:36,600 --> 00:04:45,550 So again, this network is comparing this model with the closest one which it has in its memory. 49 00:04:46,170 --> 00:04:50,190 So this is how the association neural network works. 50 00:04:50,970 --> 00:04:56,250 There are several more techniques of neural nets for each of these are the most popular one. 51 00:04:56,670 --> 00:05:03,540 And in this course, we are going to focus on classification, neural network technique.