 This sensor tile includes the Embedded Machine Learning application. Its neural network architecture is shown here with three input neurons, one hidden layer, and six output neurons. The architecture is configurable. However, this architecture will be used in the applications that follow now. We'll now demonstrate the sensor tile that learns. We start the application here through detection of a double top. And then let's execute motions. That's one, just a simple displacement from left to right. A displacement from right to left. A third motion pattern will be a displacement followed by a rotation. We'd like the sensor tile to learn another. So a displacement followed by a rotation. That's the fourth motion pattern. Let's try moving vertically. That's the fifth motion pattern. And then a sixth is a displacement and another rotation here in the vertical direction. Now the sensor tile is training its neural network. At this time, the neural network has been trained. The sensor tile is ready to recognize motion. We just executed the fifth motion pattern. And the number of blinks indicates what the sensor tile has recognized. It recognized motion five. This was our third motion. Let's count the blinks. Three blinks. Remember the sixth motion pattern? We had moved, rotated, and we'll examine what the sensor tile has identified for five, six. And just one more. Our motion pattern that was the original, the first one, one blink.