Assignment 7: Please describe how to classify red panda and white dog
Download 206.58 Kb. Pdf ko'rish
|
Assignment 7
Assignment 7: Please describe how to classify red panda and white dog. Karshiev Sanjar (ID: 2019221210) Before explaining how to classify red panda and white dog, let us talk about the general idea of the classification problem. Assume that we have 2 classes in 2D space, and each class has several data points given with red and yellow colors. Each data point has a feature vector (x1, x2). We have to train a function to learn the labels from that features. Here, the main goal is to find the optimal boundary to classify the two categories.
Now we will discuss about classifying red panda and white dog. First, we will extract red and white color from each image. The image of red panda consists of 70 % of brown color and 50 % of white color respectively. The image of white dog has 5 % of brown color and 90 % of white color. In the 2D space, we have feature data points. X axis represents white color and Y axis shows the quantity of brown color.
We can see from the above graph, the value of brown in Class1 is higher than in Class2. It means that class 1 is Red panda. Accordingly, the amount of white color is higher in Class 2 comparing with in Class 1. So, the classifier recognize this class as white dog. The classifier line is represented as w vector value. If the hypothesis of classifier, h w = w T x, is greater than zero, the classifier classifies that data point as Red Panda. If the hypothesis of classifier, h w = w T x, is smaller than zero, the classifier classifies that data point as White Dog. 𝑤
𝑥 > 0 | 𝑅𝑒𝑑 𝑃𝑎𝑛𝑑𝑎 𝑤 𝑇 𝑥 < 0 | 𝑊ℎ𝑖𝑡𝑒 𝐷𝑜𝑔
Optimal boundary Class1 Class2
Optimal boundary (classifier) w Download 206.58 Kb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling