Agents
These are signals from the environment that an agent receives through its sensors.
What are percepts?
This uninformed search method combines the best features of breadth-first and depth-first search.
What is iterative deepening?
In this type of learning, only some of the training data points are labeled.
What is semisupervised learning?
This function is applied to the weighted sum of inputs into a neuron.
What is the activation function?
This Python type is like a list, but immutable.
What is a tuple?
This term describes an agent that maximizes its expected performance measure.
What is a rational agent?
If an appropriate heuristic is available, this is the optimal search algorithm.
What is A* search?
This happens when the model memorizes the training data and fails to generalize to new data.
What is overfitting?
This term refers to network layers than are neither input nor output layers.
What are hidden layers?
This term refers to methods like __init__ and __repr__ which Python automatically calls in certain situations.
What are magic methods?
This kind of agent has no memory, but only reacts to the most recent percept.
What is a reflex agent?
Given enough time, this algorithm would play perfectly in games like Tic-Tac-Toe, Checkers, and Chess.
What is minimax search?
To avoid data snooping, this third data set is sometimes used in addition to the training and testing sets.
What is the validation set?
Independently discovered by multiple researchers in the 1980s, this technique allows gradient descent to adjust the weights of a multilayer network.
What is backpropagation?
This library allows for fast operations on arrays of numbers.
What is NumPy?
This term describes an environment where a given action does not always result in the same change.
What is nondeterministic?
This randomized search algorithm was important in beating human champions at the game of Go.
What is Monte Carlo tree search?
This technique involves growing multiple decision trees, each on a different subset of the training data.
What are random forests?
This type of unit's output equals its input if it is positive, but is zero otherwise.
What is a rectified linear unit (ReLU)?
This TensorFlow method is used to actually train a network on data.
What is fit?
This type of environment doesn't change while the agent is thinking, but the agent's performance measure does.
What is a semidynamic environment?
This term refers to the number of options available at each step in a search problem.
What is branching factor?
This type of problem involves predicting a continuous number, rather than placing an input into one of a finite set of categories.
What is regression?
This type of network is trained to output the same thing it was given as input.
What is an autoencoder?
This type of TensorFlow layer is used before the dense layer(s) at the end of a convolutional neural network.
What is Flatten?