WEBVTT 00:00.240 --> 00:02.520 -: Hello and welcome to this tutorial. 00:02.520 --> 00:04.980 So now that we have a brain and a body, 00:04.980 --> 00:06.720 we can make the AI. 00:06.720 --> 00:08.610 Of course, we're going to assemble them 00:08.610 --> 00:10.740 and you're gonna see how now things 00:10.740 --> 00:12.300 are all gonna make sense. 00:12.300 --> 00:14.070 You're gonna understand why we had 00:14.070 --> 00:15.330 to make some classes 00:15.330 --> 00:18.180 or more specifically, the purpose of making these classes, 00:18.180 --> 00:20.490 and mostly you're gonna get intuitive sense 00:20.490 --> 00:22.830 of how you can play with objects 00:22.830 --> 00:25.740 to create new objects, bigger objects. 00:25.740 --> 00:26.880 And as you understood, 00:26.880 --> 00:28.620 we already created two classes: 00:28.620 --> 00:30.240 the brain and the body. 00:30.240 --> 00:33.090 So then by creating some objects of these two classes, 00:33.090 --> 00:35.100 well, we can make a bigger object, 00:35.100 --> 00:36.480 which will be the AI. 00:36.480 --> 00:38.730 And that's exactly what we're gonna do now. 00:38.730 --> 00:42.120 We are gonna make the AI by assembling the two classes 00:42.120 --> 00:45.030 that we created, that is the brain and the body. 00:45.030 --> 00:46.170 So let's do this. 00:46.170 --> 00:48.390 It's gonna become super intuitive now. 00:48.390 --> 00:51.540 I'm going to introduce a new class, 00:51.540 --> 00:54.090 which of course, I'm gonna call AI 00:54.090 --> 00:57.240 because now we're ready to make the final AI. 00:57.240 --> 01:02.240 And right here, I'm going to start with the init function. 01:04.530 --> 01:07.350 And what is this init function going to take? 01:07.350 --> 01:10.320 It is right now that things will make a lot of sense. 01:10.320 --> 01:13.500 Well, first, we get the inevitable self 01:13.500 --> 01:14.580 with the object. 01:14.580 --> 01:17.970 But then, the two arguments are gonna be nothing else 01:17.970 --> 01:21.390 than the brain and the body. 01:21.390 --> 01:25.230 That's the two elements that we need to build an AI. 01:25.230 --> 01:27.390 We need a brain, the neural network 01:27.390 --> 01:30.600 and a body, which will play the action with softmax. 01:30.600 --> 01:33.000 And so now very simply what we have to do 01:33.000 --> 01:35.610 is defining the variables of our AI object 01:35.610 --> 01:38.730 because right now, brain and body are just arguments. 01:38.730 --> 01:39.660 So let's do this. 01:39.660 --> 01:41.490 We have to make two variables obviously. 01:41.490 --> 01:43.950 The first one is self.brain, 01:43.950 --> 01:47.190 which will be equal to the brain argument 01:47.190 --> 01:50.940 but which later will be a brain object created 01:50.940 --> 01:52.710 from the CNN class. 01:52.710 --> 01:56.730 And then the second variable is self.body, 01:56.730 --> 02:00.660 which will be equal to the body argument right here. 02:00.660 --> 02:02.580 But which in the future will be, 02:02.580 --> 02:06.150 of course, an object of the softmax body class. 02:06.150 --> 02:08.100 And so now as simple as that, 02:08.100 --> 02:11.280 thanks to this pretty awesome structure that we made here, 02:11.280 --> 02:12.780 we built an AI. 02:12.780 --> 02:15.000 Not trained yet, not intelligent yet 02:15.000 --> 02:17.550 but still, we have a brain and a body. 02:17.550 --> 02:20.160 So that's perfect, and now the only thing 02:20.160 --> 02:21.480 that we have to do left 02:21.480 --> 02:24.660 is make a big forward function 02:24.660 --> 02:26.910 that will do the whole propagation. 02:26.910 --> 02:29.130 Since now our AI is assembled, 02:29.130 --> 02:30.930 well, we're not gonna use separately 02:30.930 --> 02:32.580 the two forward functions, 02:32.580 --> 02:34.950 the one of the brain and the one of the body. 02:34.950 --> 02:37.140 We're gonna make a big forward function, 02:37.140 --> 02:39.090 which will be our next function, 02:39.090 --> 02:42.630 which will take the images as input, 02:42.630 --> 02:45.360 then we'll propagate the signals in the brain, 02:45.360 --> 02:46.410 and for this, we will, of course, 02:46.410 --> 02:48.780 use the first forward function, 02:48.780 --> 02:51.840 and then once we get the output signals of the brain, 02:51.840 --> 02:54.990 we will forward these output signals into the body 02:54.990 --> 02:56.160 with the forward function here 02:56.160 --> 02:58.320 that will use a softmax technique. 02:58.320 --> 03:01.530 And then eventually, we will return the actions to play. 03:01.530 --> 03:02.520 And that is just now 03:02.520 --> 03:04.110 that we can make this big forward function 03:04.110 --> 03:07.800 because we have assembled the brain and the body. 03:07.800 --> 03:10.650 And so we're not gonna call this next function forward, 03:10.650 --> 03:12.450 we're gonna use the call function, 03:12.450 --> 03:13.920 which is kind of like the init function 03:13.920 --> 03:17.070 but that will call the two forward functions here 03:17.070 --> 03:20.190 from the brain and the body to propagate the signal 03:20.190 --> 03:22.560 from the very beginning with the input images 03:22.560 --> 03:25.170 to the very end with the action to play. 03:25.170 --> 03:28.290 And that's why the next function we will make now 03:28.290 --> 03:30.390 is this call function 03:30.390 --> 03:33.090 that will combine the two forward functions 03:33.090 --> 03:34.830 of the brain and the body. 03:34.830 --> 03:36.360 That's the only thing we have to do 03:36.360 --> 03:39.480 and then we will have our AI ready. 03:39.480 --> 03:42.150 So let's make this final step in the next tutorial. 03:42.150 --> 03:43.953 And until then, enjoy AI.