TensorFlow Basics and Tensors

# CHAPTER 6

TensorFlow Basics and Tensors #

1. Introduction #

2. Learning Objectives #

• Define what a Tensor is.

• Create TensorFlow Constants and Variables.

• Perform mathematical operations on Tensors.

• Check Tensor shapes and data types.

• Explain the concept of the Computation Graph.

3. What are Tensors? #

1. 1. Tensors can be processed on a GPU (Graphics Processing Unit) for massive speedups.

1. 2. Tensors automatically keep track of the mathematical operations applied to them, allowing TensorFlow to automatically calculate calculus gradients for Backpropagation (known as AutoGraph).

Tensor Dimensions (Ranks):

• 0D Tensor (Scalar): A single number. (e.g., 5)

• 1D Tensor (Vector): An array of numbers. (e.g., [1, 2, 3])

• 2D Tensor (Matrix): A grid of numbers. (e.g., A grayscale image).

• 3D Tensor: A cube of numbers. (e.g., A color image with Red, Green, Blue channels).

• 4D Tensor: A batch of color images.

4. TensorFlow Constants #

import tensorflow as tf

# Creating a 0D Tensor (Scalar)
scalar = tf.constant(42)
print(f"Scalar: {scalar}")

# Creating a 1D Tensor (Vector)
vector = tf.constant([1.0, 2.0, 3.0])
print(f"Vector Shape: {vector.shape}")

# Creating a 2D Tensor (Matrix)
matrix = tf.constant([
    [10, 20],
    [30, 40]
])
print(f"Matrix Dimensions: {matrix.ndim}") # Output: 2

5. TensorFlow Variables #

# Create a variable
my_weights = tf.Variable([0.5, 0.8])

# Change the value of the variable (You cannot do this with a constant!)
my_weights.assign([0.6, 0.9])

print(my_weights)

6. Tensor Operations #

tensor_a = tf.constant([1, 2, 3])
tensor_b = tf.constant([4, 5, 6])

# Addition
added = tf.add(tensor_a, tensor_b) # or simply: tensor_a + tensor_b

# Multiplication
multiplied = tf.multiply(tensor_a, tensor_b)

# Matrix Multiplication (Crucial for Neural Networks)
# Note: Matrix multiplication requires specific shapes!
mat_a = tf.constant([[1, 2], [3, 4]])
mat_b = tf.constant([[5, 6], [7, 8]])
dot_product = tf.matmul(mat_a, mat_b)

print("Dot Product:\n", dot_product)

7. Mini Project: Tensor Operations Playground #

import tensorflow as tf

# 1. Create a tensor filled with random numbers from a normal distribution
# Shape: (3, 3) - A 3x3 matrix
random_tensor = tf.random.normal(shape=(3, 3), mean=0.0, stddev=1.0)

print("Random Tensor:\n", random_tensor)

# 2. Find the maximum value in the tensor
max_val = tf.reduce_max(random_tensor)
print("\nMax Value:", max_val.numpy())

# 3. Find the index of the maximum value
max_index = tf.argmax(random_tensor[0]) # Checking the first row
print("Index of max value in row 1:", max_index.numpy())

*Notice .numpy()? If you just want the raw Python number out of a complex Tensor object, call .numpy()!*

8. The Computation Graph #

9. Common Mistakes #

• Shape Mismatch in Matrix Multiplication: You cannot use tf.matmul on two tensors unless the inner dimensions match. E.g., A (3, 2) matrix can multiply with a (2, 5) matrix, but NOT a (3, 5) matrix. This is a strict rule of linear algebra, not just TensorFlow!

• Mixing Data Types: Trying to add an integer tensor [1, 2] to a float tensor [1.0, 2.0] will crash. TensorFlow is strictly typed. You must cast them using tf.cast(tensor, dtype=tf.float32).

10. Best Practices #

• Use tf.float32: Deep learning relies on decimals (gradients, weights). Ensure your input tensors are usually float32. Using float64 takes up twice as much RAM on your GPU and rarely improves accuracy.

11. Exercises #

1. 1. Create a tf.constant 2D matrix filled entirely with the number 1 (Hint: research tf.ones).

1. 2. Write code to cast an integer tensor tf.constant([10, 20]) into a float tensor.

12. MCQ Quiz with Answers #

13. Interview Questions #

• Q: Explain what "Rank" means in the context of a Tensor and give an example of a Rank-2 Tensor.

• Q: Explain what an Eager Execution environment is compared to a static Graph execution environment.

14. FAQs #

15. Summary #

16. Next Chapter Recommendation #