Which of the following are smooth, non-linear activation functions used in neural network hidden layers?
Sigmoid, Tanh, and ReLU are all standard non-linear activation functions utilized in neural networks.
What architectural innovation distinguishes the 'YOLO' (You Only Look Once) object detection model introduced by Joseph Redmon et al.?
Unlike multi-stage region proposal detectors (like R-CNN), YOLO processes the whole image in one forward pass through a single network, achieving real-time inference speeds.
What is the primary memory advantage of Depth-First Search (DFS) over Breadth-First Search (BFS)?
DFS only stores the active branch path plus unexpanded sibling nodes, requiring O(b*m) space, whereas BFS requires $O(b^d)$ exponential space.
What defines a 'Non-Monotonic Logic' system in Artificial Intelligence?
In non-monotonic logic (e.g., default logic, circumscription), acquiring new evidence can defeat existing default assumptions (e.g., learning Tweety is a penguin defeats Tweety flies).
What is the mathematical definition of the Rectified Linear Unit (ReLU) activation function?
The ReLU activation function outputs z if z > 0, and 0 otherwise: f(z) = max(0, z). Its constant derivative of 1 for positive inputs mitigates vanishing gradients.