Convolution operation concept in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a convolution operation in the context of machine learning?

A convolution operation is a way to process data by sliding a small filter (kernel) over input data (like an image) to extract important features such as edges or textures.

beginner

What does a kernel (filter) do in a convolution operation?

A kernel is a small matrix that moves over the input data and multiplies its values with the input values to highlight specific patterns or features.

intermediate

How does stride affect the convolution operation?

Stride controls how many steps the kernel moves each time. A stride of 1 moves the kernel one pixel at a time, while a larger stride skips pixels, making the output smaller.

intermediate

What is padding in convolution and why is it used?

Padding adds extra pixels (usually zeros) around the input edges to control the output size and help the kernel cover edge pixels properly.

beginner

In TensorFlow, which function is commonly used to perform a 2D convolution?

The function tf.nn.conv2d is used to perform 2D convolution on input data like images in TensorFlow.

What does the kernel do during convolution?

A Removes pixels from the input

B Adds noise to the input data

C Slides over input and multiplies values to detect features

D Randomly changes input values

The kernel slides over the input and multiplies its values with input values to detect features.

What happens if you increase the stride in a convolution operation?

A Output size decreases

B Output size increases

C Kernel size increases

D Padding is removed

Increasing stride makes the kernel jump more pixels, reducing the output size.

Why do we use padding in convolution?

A To keep output size same as input

B To reduce kernel size

C To speed up training

D To add color to images

Padding adds pixels around input edges to keep output size the same or control it.

Which TensorFlow function performs 2D convolution?

A tf.matmul

B tf.nn.conv2d

C tf.reduce_sum

D tf.reshape

tf.nn.conv2d is the function used for 2D convolution in TensorFlow.

What is the main purpose of convolution in image processing?

A To remove colors

B To increase image size

C To convert images to text

D To extract important features like edges

Convolution helps extract features such as edges and textures from images.

Explain how a convolution operation works on an image using a kernel.

Think about how a small window moves over a picture to find patterns.

Describe the roles of stride and padding in convolution operations.

Consider how moving faster or adding borders changes the result.