Pattern recognition

Question 3: Recall that in your take-home assignment you were asked to devise a method to allow for each class to have multiple representative embedding (or weight vectors) in the transformation matrix. You will be required to do a similar task here but for the binary classification task. Suppose you have a Model m that outputs an embedding vector e. In the normal binary classification task, e is transformed into a score s by multiplying with weight matrix W, which has a single row vector. The score s is then passed through the sigmoid activation function and the resulting value (p) is treated as a probability and used in a binary-cross-entropy loss function. In a new setting, the weight matrix W will have 4 row vectors, resulting in 4 scores (s, -5.). The probability value p is calculated as follows: p = sigmoid (max (S1, S2, S3, S4)) Your task is to simply modify the binary-cross-entropy loss function to accommodate the effect of the extra row vectors. Question 3: Recall that in your take-home assignment you were asked to devise a method to allow for each class to have multiple representative embedding (or weight vectors) in the transformation matrix. You will be required to do a similar task here but for the binary classification task. Suppose you have a Model m that outputs an embedding vector e. In the normal binary classification task, e is transformed into a score s by multiplying with weight matrix W, which has a single row vector. The score s is then passed through the sigmoid activation function and the resulting value (p) is treated as a probability and used in a binary-cross-entropy loss function. In a new setting, the weight matrix W will have 4 row vectors, resulting in 4 scores (s, -5.). The probability value p is calculated as follows: p = sigmoid (max (S1, S2, S3, S4)) Your task is to simply modify the binary-cross-entropy loss function to accommodate the effect of the extra row vectors