Hands-On Transfer Learning with Python

Tensor operations

All computation done for training/testing a deep neural network can be formulated by a set of tensor operations. For example, adding, multiplying, and subtracting tensors. Following are some of the frequently used tensor operations in this book:

• Element-wise operations: Applying a function independently to all elements of a tensor is very commonly used in deep learning. For example, applying an activation function to all the units in a layer. Other element-wise operations include applying basic math operators, such as +, -, and *, element wise to two tensors of identical shape.
• Tensor dot: The dot product of two tensors is different from the elementwise product of two tensors. The dot product of two vectors is a scalar ...