Because Prof Ng has defined the layout of the W matrix so that a transpose is no longer required. It is required in the Logistic Regression case only because he chooses to define the weight vector w as a column vector. Here’s another recent post that talks about this in more detail.