1. A multi-layer perceptron is used to decide the execution of one of the following actions in an assembly production line: Discard rotten pears, which look brown. Keep on storing green pears. Place yellow pears in a container so that they will be distributed into the market to be sold. The recognition mechanism is based on a small camera, which captures a very small image (known as image patch) of a pear as shown in the following Figure. This image patch is modeled as a 9 x 9 matrix of pixels with each pixel having three RGB components. The extracted information (all RGB components of the pixels on the image patch) is stretched out as a feature vector, and it is fed to the multi-layer perceptron. Assume that the multi-layer perceptron is built by using one hidden layer with 20 neurons. Determine the number of output neurons to use and number of inputs to each neuron at the first hidden layer. Draw the MLP architecture. TXIR | XG X1B X= : |XpG [