Part 5 - Support Vector Machines (5P) In the toy dataset (see Figure 5), linearly separable...

Transcribed Image Text:

Part 5 - Support Vector Machines (5P) In the toy dataset (see Figure 5), linearly separable training data points a € R² are given. feature x 6 4 2 0 2 9 -6 feature X class (-1) class (+1) 4 Figure 5: Toy dataset for SVM. (a) (1P) Highlight all support vectors for the optimal linear SVM solution in Fig. 5. (b) (2P) What is the leave-one-out cross-validation error (provide the number!) that you would get on the given data set in Fig. 5? Which data point(s) cause this error? Why? (c) (1P) SVM is an instance-based model. What does this mean? Give exactly one example of a non-instance based model class. (d) (1P) Considering the dual formulation of a trained SVM model, that shall be deployed on a device with very little storage: Which parameters of this formulation determine the minimal set of training data points that must be stored? Part 5 - Support Vector Machines (5P) In the toy dataset (see Figure 5), linearly separable training data points a € R² are given. feature x 6 4 2 0 2 9 -6 feature X class (-1) class (+1) 4 Figure 5: Toy dataset for SVM. (a) (1P) Highlight all support vectors for the optimal linear SVM solution in Fig. 5. (b) (2P) What is the leave-one-out cross-validation error (provide the number!) that you would get on the given data set in Fig. 5? Which data point(s) cause this error? Why? (c) (1P) SVM is an instance-based model. What does this mean? Give exactly one example of a non-instance based model class. (d) (1P) Considering the dual formulation of a trained SVM model, that shall be deployed on a device with very little storage: Which parameters of this formulation determine the minimal set of training data points that must be stored?

Answer rating: 100% (QA)

4 vectors for the optimal linear SVM solution V V J b a 12 1 V 024... View the full answer