So, basically, in NN, whole layers are abstracted as Matrixes, and most used function to simulate NNs is Matrix multiplication. That actually explains why NN systems are easily accelerated by GPUs — the 3D visualisation is heavily based on matrix algebra, too! How to we train them?
NumPy Matrix Multiplication You've heard a lot about matrix multiplication in the last few videos – now you'll get to see how to do it with NumPy. However, it's important to know that NumPy supports several types of matrix multiplication. Element-wise Multiplication You saw some element-wis...