In the handwritten digit classification example, we used the average of a 3x3 center of each image as an attribute and developed a thresholding classifier that classified any image with the attribute greater than the threshold as digit 1 and otherwise as digit O. The classification accuracy of 1's and O's as well as the total accuracy per threshold value are shown in the graph below. 1. What threshold value should be selected to give an optimal classification result? Explain your answer. 2. What classification accuracy this model will provide for detecting O's and what classification accuracy for 1's? Explain. 3. What is the total classification accuracy? Explain. Zero ano un cor, ungen 100 30 80- 70 00 Percent correct IF center > threshold digit=1 ELSE digit=0 END 50 40 - Conectos % Correct 15 -Overal Accuracy 30 20 10 81 100 200 250 300 150 Threshold value In the handwritten digit classification example, we used the average of a 3x3 center of each image as an attribute and developed a thresholding classifier that classified any image with the attribute greater than the threshold as digit 1 and otherwise as digit O. The classification accuracy of 1's and O's as well as the total accuracy per threshold value are shown in the graph below. 1. What threshold value should be selected to give an optimal classification result? Explain your answer. 2. What classification accuracy this model will provide for detecting O's and what classification accuracy for 1's? Explain. 3. What is the total classification accuracy? Explain. Zero ano un cor, ungen 100 30 80- 70 00 Percent correct IF center > threshold digit=1 ELSE digit=0 END 50 40 - Conectos % Correct 15 -Overal Accuracy 30 20 10 81 100 200 250 300 150 Threshold value