Snek fren
Unlike C and Java, Python doesn't require you to declare these
What are types?
This adversarial search technique can be used for games where one player tries to make a score as high as possible, while the other tries to reduce it
What is minimax search?
In unsupervised learning, the training data lack these
What are labels (or correct answers)?
Each internal node of a decision tree has this many children
What is two?
Examples of this include the Heaviside threshold function, the logistic sigmoid function, and the rectified linear unit function
What is an activation function?
Python uses this instead of curly braces to indicate a block of code
What is indentation?
If it is not possible to search all the way to the end of a game, this kind of function is needed
What is a static evaluation function?
This unsupervised learning application find data points that are very different from the rest of the data
What is anomaly detection?
In a decision tree that performs perfectly on the training set, this is true of the data points associate with a given leaf
What is all being of the same class?
A single neuron famously cannot compute this simple Boolean function
What is XOR?
This Python data structure is similar to a list, but immutable
What is a tuple?
This search improvement finds the same answer with less computation by ruling out branches that are too good to be true
What is alpha-beta pruning?
In k-means clustering, k refers to this
What is the number of clusters?
This would happen if a decision tree used no regularization at all
What is overfitting?
This technique, independently discovered by several researchers, allows for gradient descent to train multi-layer neural networks
What is backpropagation?
Using colons between square brackets, this Python technique allows you extract part of a list, string, or other sequence
What is slicing?
This adversarial search technique involves completing a game randomly many times
What is Monte Carlo tree search?
K-means clustering alternates between updating the positions of the centroids and doing this with each data point
What is associating it with the nearest centroid?
This term refers to a group of predictors that vote on an answer for each data point
What is an ensemble?
Early attempts at training deep neural networks were thwarted by this problem, where a change in a weight in an early layer would have almost no effect on the network's output
What is the vanishing gradient problem?
What is a first-class object?
This program was the first to beat top human players at the classical Asian game of Go
What is AlphaGo?
This refers to the situation where only a small fraction of the training instances are labeled
What is semi-supervised learning?
This technique involves training several decision trees on different samples of the same data
What is a random forest?
This technique is used to apply the same weights to different receptive fields across an image
What is convolution?