The following shows a table where the input attributes are x1 and x2 and y is the target attribute. There are two possible values in the target attribute y, namely "Yes" and "No". Attribute ID corresponds to the ID of each tuple. (a) Consider a classification task by a neural network model or SVM. In order to perform the classification task by a neural network model or SVM, we need to transform the above table T to another table T ' such that the target attribute y in the transformed table contains numeric values (i.e., 1 for "Yes" and -1 for "No") instead of categorical values. What is this transformed table T'? (b) Consider a classification task by SVM. Formulate this SVM problem by a quadratic programming. Please list the objective function and all constraints in this quadratic programming. (c) Consider a classification task by a neural network model containing a single neuron. Initially, we set the values of w1,w2 and b to be 0.1 where b is a bias value in the neuron. Suppose the learning rate is denoted by . Let =0.5. Suppose we adopt the hyperbolic tangent function as an activation function. Please try to train the neural network with the following first six instances in the given sequence and the answer in Part (a). 1. (x1,x2)=(13,13) 2. (x1,x2)=(19,9) 3. (x1,x2)=(21,15) 4. (x1,x2)=(15,19) 5. (x1,x2)=(7,9) 6. (x1,x2)=(5,7) What are the values of w1,w2 and b after each of these six instances is processed? Please show each answer up to 4 decimal places. (d) What is the major disadvantage of the traditional neural network model compared with the recurrent neural network model