Appearance
Artificial Intelligence - Glossary
| Term | Simple meaning |
|---|---|
| A* | Search using cost paid plus estimated cost remaining |
| Activation function | Function converting a neuron's input into output |
| Actuator | Component through which an agent acts |
| Admissible heuristic | Estimate that never exceeds true remaining cost |
| Agent | System that perceives and acts in an environment |
| Alpha-beta pruning | Removes game-tree branches that cannot affect minimax |
| Backpropagation | Computes neural-network gradients from output toward input |
| Bayes' rule | Updates a hypothesis probability using evidence |
| Bayesian network | Directed acyclic graph of probabilistic dependencies |
| Bias | Adjustable constant added to a neuron's weighted sum |
| Breadth-first search | Expands the shallowest frontier node first |
| Conditional independence | Independence after another variable is known |
| Consistent heuristic | Heuristic obeying the triangle-style cost condition |
| Constraint satisfaction problem | Variables, domains, and constraints |
| Depth-first search | Expands the deepest frontier node first |
| Environment | External world in which an agent operates |
| Frame | Slot-value structure describing an object or situation |
| Gradient descent | Moves parameters opposite the loss gradient |
| Heuristic | Estimate used to guide search |
| Hidden Markov model | Temporal model with hidden states and observations |
| Knowledge graph | Network of entities and labelled relationships |
| Learning rate | Step size used during parameter updates |
| Loss function | Number measuring prediction error |
| Minimax | Chooses a move assuming an optimal opponent |
| Naive Bayes | Classifier assuming conditional feature independence |
| Ontology | Formal vocabulary of domain concepts and relationships |
| PEAS | Performance, Environment, Actuators, Sensors |
| Perceptron | Linear threshold learning unit |
| Rational agent | Agent choosing actions expected to maximize performance |
| Semantic network | Graph representation of concepts and relations |
| Sensor | Component through which an agent perceives |
| State space | Set of states reachable in a problem |
| Utility | Numerical preference for an outcome |
| Uniform-cost search | Expands the frontier node with lowest path cost |