A linear regression model will use this shape to fit the dataset.
What is a line?
Neural networks are based on this brain structure/cell.
What are neurons?
Stores a value and is assigned a name.
What is a variable?
These make up any image, where they have values between 0 and 1 for black-and-white images and three unique values for colored images.
What are pixels?
The numerical difference between predicted output and actual output, which allows the programmer to see how well our model is performing.
What is error?
A variable with a numerical value that a linear regression will aim to predict.
What is a continuous variable?
Neural networks apply this process to find the best parameters for a model with the lowest error.
What is gradient descent?
This is a whole number.
What is an integer?
Process to of altering the original data in order to make new copies that have slight variations from the original. This makes the model more robust to unseen data.
What is data augmentation?
We can say our model is ______ when training loss is really high and generalization/testing loss is really high.
What is underfitting?
Used in the linear regression formula, commonly referred to as rise-over-run.
What is slope?
The values stored in a neural network that allows our model to determine which inputs are more important than others.
What are weights?
This is a set of characters surrounded by quotations.
What is a string?
We use this split after training our model in order to see how it performs on data it hasn't seen before.
What is a testing split?
We can say our model is ______ when training loss is really low and generalization/testing loss is really high.
What is overfitting?
The relationship between two continuous variables (doesn't necessarily mean that one causes the other).
What is correlation?
Special type of neural network that is super useful for image processing and prediction which uses filters to find patterns in images.
What are convolutional neural networks (CNNs)?
A way to condense code that can be reused with different parameters.
What is a function?
After collecting our data points, we do this to filter through the dataset to ensure that our dataset only has relevant and high quality data in it.
What is cleaning the dataset?
When the model makes implicit assumptions. If deployed, the model could risk discriminating against certain demographics.
What is bias?
Each example in our dataset is associated with a _____ that tells us the true output value.
What is a label?
The special layers between the input and output layers that find relationships between the data and do all the math for the programmer.
What are hidden layers?
A data structure that can hold multiple values, including strings, integers and more!
What is a list?
A special split from the training split that is used to tell the model how it is performing during training.
What is a validation split?
Our classifier fails to recognize certain groups. This is because our model needs to be trained on a _____ and _____ dataset.
What is a large and high quality dataset?