Deep Learning Flashcards
Precision
The fraction of positives that are true positives
Recall
The fraction of true positives that were flagged as positives
Formula for F-score
2(precision * recall) / (precision + recall)
In convolution, what is the stride value?
How many pixels the filter is moved to the right after each step
What does a pooling layer do?
Reduces the size of the image to speed up computation
What is semantic segmentation?
When every pixel in an image is categorised into a class
What is instance segmentation?
Identifying and separating individual objects within an image
What is panoptic segmentation?
Combines semantic segmentation and instance segmentation to provide a comprehensive understanding of the scene. In a given image, every pixel is assigned a semantic label, and pixels belonging to “things” classes (countable objects with instances, like cars and people) are assigned unique instance IDs.
What are the two types of supervised learning
Regression and classification
What is a fully connected layer?
When a layer in a neural network takes every possible parameter as input. In computer vision, that would mean every pixel in an image.
What is a locally connected layer?
When a layer in a neural network takes in only part of the total parameters as input. In computer vision, that would mean an area in an image.
List 3 uses for deep learning in computer vision
Classification
Segmentation
Detection
What is a loss function?
A function that maps an event or values of one or more variables onto a real number intuitively representing some “cost” associated with the event.
What is an activation function?
A node in a neural network. Calculates the output of the node based on the inputs and weights.
List 4 types of activation function
Identity
Threshold
ReLU (Rectified Linear Unit)
Sigmoid
