TENSORS…
--
~Hola folks!!! Nice to be back with another post 😊…
Tensor is a data structure. It is a container where 90% are numbers. There are different types of tensors.
Before look what are types of tensors, Let’s look what is Rank, Shape and Size of a Tensor.
- Rank : No. of Axes = No. of dimensions (ndim) = Rank
- Shape : How many values can be stored in an axis
- Size : Available elements in the shape
Let’s move onto Types of Tensors…
0-D Tensors/ Scalers
Scalers or the 0-D (0 Dimension) tensors are the tensors where a single number is stored. It has no direction.
ex: 2
Following code snippet can be used to check the dimension of a Scaler…
import numpy as np
a = np.array(2)
a
a.ndim#output => 0
#size = 1
1-D Tensors/ Vectors
Vectors or the 1-D (1 Dimension) tensors are a collection of scalers. It includes list of values.
ex: [1,2,3,4]
Following code snippet can be used to check the dimension of a Vector…
import numpy as np
b = np.array([1,2,3,4])
b
b.ndim#output => 1
#size = 4
Vector dimension gives how many values in the array.
2-D Tensors/ Matrices
Matrices or the 2-D (2 Dimension) tensors are a collection of multiple vectors. I.E it includes collection of lists.
ex: [1,2,3][4,5,6][7,8,9]
Following code snippet can be used to check the dimension of a Matrix…
import numpy as np
c = np.array([[1,2,3],[4,5,6],[7,8,9]])
c
c.ndim#output => 2
#shape = (3,3)
#size = 9
3-D Tensors/ N-D Tensors/ Cuboids
3-D (3 Dimension) tensors are a collection of matrices. It includes row number, column numbers and a depth i.e. number of matrices.
ex: [ [[1,0,0,0][0,1,0,0]], [[1,0,0,0][0,0,1,0]] ,[[1,0,0,0][0,0,0,1]]] =>
row # * column # * depth = 3 * 2 * 4shape = (3,2,4)
size = 3*2*4 = 24
4-D Tensors
Collection of cuboids/ 3-D Tensors is called 4-D Tensor. We can find 4-D tensors in the computer vision domain.
ex: Batch of RGB images
5-D Tensors
Collection of 4-D Tensors is called 5-D Tensor.
ex: Batch of videos (collection of frames)
Conclusion
We just briefly learned about Tensors…
That’s it for now, and please press the clap button if you find this article helpful.
This might be a continuation of a series of Machine Learning Fundamentals.🤞
Special thanks to @Jinendra Bogahawatte
