Glossary for *Artificial Intelligence: A Modern Approach*

# Glossary for *Artificial Intelligence: A Modern Approach*

[A](#a) | [B](#b) | [C](#c) | [D](#d) | [E](#e) | [F](#f) | [G](#g) | [H](#h) | [I](#i) | [J](#j) | [K](#k) | [L](#l) | [M](#m) | [N](#n) | [O](#o) | [P](#p) | [Q](#q) | [R](#r) | [S](#s) | [T](#t) | [U](#u) | [V](#v) | [W](#w) | [X](#x) | [Y](#y) | [Z](#z)

## 8-puzzle

**8-puzzle** consists of a 3x3 grid containing 8 numbered tiles and a blank space. A tile adjacent to the blank space can slide into that space. The object is to reach a specified **goal state** from a given **initial state**.

# A

## absolute error

Magnitude of the difference between the theoretical value (expected value) and the actual value of a physical quantity.

## abstraction

Abstraction is selecting data from a larger pool to show only the relevant details to the object.

## abstraction hierarchy

It hides the complexity of the system and allows individuals to work on different modules of the hierarchy at the same time.

## accessibility relations

In modal logic, an accessibility relation R is a binary relation such that R⊆ W×W where W is a set of possible worlds. The accessibility relation determines for each world w ∈ W which worlds ẃ are accessible from w.

## action monitoring

Checking the preconditions of each action as it is executed, rather than checking the preconditions of the entire remaining plan.

## action schema

## action-utility function

## actions

The things that an agent can do. We model this with a function, **Actions(s)**, that returns a collection of actions
that the agent can execute in state *s*.

## activation

## activation function

A mathematical function that transforms the input or set of inputs received at a neuron to produce an output.
Popular examples include the Sigmoid function, Rectificied Linear Units (ReLU) and Hyperbolic Tangent (Tanh)

## active learning

An active learning agent decided which actions to take in order to guide its learning: it values leearning new things
as well as reaping immediate rewards from the environment.
This is in contrast to a passive learning agent, which learns from its observations, but the actions the agent takes are not influenced by the learning process.

## actor

## adaptive dynamic programming

Also known as Approximate Dynamic Programming; it is a type of Reinforcement Learning where local rewards and transitions depend on unknown parameters - we set an initial control policy and update it until it converges to an optimal control policy.

## add list

## admissible heuristic

A **heuristic** is a function that scores alternatives at each branching in a search algorithm. An **admissible heuristic** is one that *never overestimates* the cost to reach the goal. Admissible heuristics are **optimistic** in nature as they believe the cost of reaching the goal is less than it actually is.

## adversarial search

Traversing a tree data structure to find all possible moves. It is usually used in a two-player game; each available move is represented using gain and loss for an individual player. An important application of it is in zero sum games, as in those games, one players' loss is the other players' gain.

## adversary argument

## agent

An **agent** is anything that can be viewed as perceiving its **environment** through **sensors** and acting upon that environment through **actuators**.

## agent function

An agent's behavior is described by the **agent function** that maps any given percept sequence to an action.

## agent program

_Internally,_ the agent function for an artificial agent will be implemented by an **agent program**.

## agglomerative clustering

It is a category of hierarchical clustering which uses a bottom-up approach. All observations start in their own cluster and different pairs of clusters are merged as you move up levels in the hierarchy. Its results are represented using a dendrogram.

## aggregation

## algorithm

An **algorithm** is a sequence of **unambiguous finite steps** that when carried out on a given problem produce the expected outcome and terminate in **finite time**.

## alignment method

## alpha-beta

**alpha** (**α**) is the value of the best (i.e., highest-value) choice we have found so far at any choice point along the path for MAX and **beta**(**β**) is the value of the best (i.e., lowest-value) choice we have found so far at any choice point along the path for MIN in a standard minimax tree.

## alpha-beta pruning

**alpha—beta pruning** is applied to a standard minimax tree to prune away branches that cannot possibly influence the final minimax decision.

## ambient illumination

Light that is already present in a scene, before any additional lighting is added. It usually refers to natural light.

## ambiguity

It is the state of being uncertain or doubtful.

## ambiguity aversion

Ambiguity aversion (also known as uncertainty aversion) is a preference for known risks over unknown risks.

## analogical reasoning

Analogical reasoning is any type of thinking that relies upon an analogy. An analogical argument is an explicit representation of a form of analogical reasoning that cites accepted similarities between two systems to support the conclusion that some further similarity exists.

## anchoring effect

It is a type of cognitive bias which makes people focus on the first piece of information (the "anchor") that was given to them, to make decisions. To explain this with an example; when buying a product if you're told a high price by the seller, your mind estimates the worth of that product as that initial/anchor price you're told, and then when you're offered a discount on it, you are more inclined to buy it thinking that you're getting it for cheap.

## And-Elimination

In propositional logic, conjunction elimination (also called and elimination, ∧ elimination, or simplification0 is a valid immediate inference, argument form and rule of inference which makes the inference that, if the conjunction A and B is true, then A is true, and B is true. The rule makes it possible to shorten longer proofs by deriving one of the conjuncts of a conjunction on a line by itself.

## AND-parallelism

## angelic nondeterminism

A notional ability always to choose the most favorable option, in constant time. With angelic non-determinism, any problem in NP would be solvable in polynomial time.

## angelic semantics

## answer literal

## answer set programming

Answer set programming (ASP) is a form of declarative programming oriented towards difficult (primarily NP-hard) search problems. It is based on the stable model (answer set) semantics of logic programming.

## answer sets

## aortic coarctation

It is a Medical term which means the narrowing of aorta - the largest artery in the body which starts from the heart.

## apprenticeship learning

It is the process of learning by observing the demonstration of an expert. In a way, it is a form of supervised learning where the training data would be the tasks performed by the expert.

## architecture

It is a very broad term; however, it is generally used to refer to the structure of the buildings and other constructions.

## arity

In logic, mathematics, and computer science, the arity of a function or operation is the number of arguments or operands that the function takes.

## artificial life

Artificial life (often abbreviated ALife or A-Life) is a field of study wherein researchers examine systems related to natural life, its processes, and its evolution, through the use of simulations with computer models, robotics, and biochemistry.

## ascending-bid

Bidders place bids of progressively higher amounts, aiming to outbid each other. The bidder who places the highest bid by the end of the auction wins.

## Asilomar Principles
The Asilomar Conference on Beneficial AI was a conference organized by the Future of Life Institute, held January 5-8, 2017, at the Asilomar Conference Grounds in California. More than 100 thought leaders and researches in economics, law, ethics, and philosophy met at the conference, to address and formulate principles of beneficial AI. Its outcome was the creation of a set of guidelines for AI research – the 23 Asilomar AI Principles.

## assignment

## associative memory

In terms of Psychology, it is the type of memory which allows us to remember things by finding links between, apparently, unrelated things. To explain with an example; remembering someones' name by the dress they wore the first time you met them. Clearly, these two things seem completely unrelated.

## asymptotic analysis

It is a Mathematical method of describing limiting behavior by using an input bound function, which means that the algorithm would run in constant time if no output was given, as the rest of the factors contributing to the computation are constant.

## asymptotic bounded optimality

## ATMS

Assumption-Based Truth Maintenance System (ATMS) allows to maintain and reason with a number of simultaneous, possibly incompatible, current sets of assumption.

## atom

An atom is the smallest constituent unit of ordinary matter that has the properties of a chemical element.

## atomic representation

## atomic sentence

## attribute-based extraction

## augmented grammar

Any grammar whose productions are augmented with conditions expressed using features.

## authority

## automatic assembly sequencing

## autonomy

It is the character of being independent and self-governing in vital as well as non-vital situaltions.

## average reward

## axiom

In Mathematics or logics, an axiom is a statement or a proposition which is assumed to be true to serve as a starting point for further arguments and reasoning. Example of an axiom: “Nothing can both be and not be at the same time and in the same respect”.

# B

## back-propagation

**back-propagation** is an algorithm used for *supervised learning* of **artificial neural networks** using gradient descent.
The method calculates the gradient of a given error function with respect to the weights of the network. The "backward" terminology stems because the gradient calculation requires backward propagation through the newtork.

## backed-up value
## backgammon

Backgammon is one of the oldest known board games. It is a two player game where each player has fifteen pieces (checkers) which move between twenty-four triangles (points) according to the roll of two dice. The objective of the game is to be first to bear off, i.e. move all fifteen checkers off the board.

## background subtraction

Foreground detection is one of the major tasks in the field of computer vision and image processing whose aim is to detect changes in image sequences. Background subtraction is any technique which allows an image's foreground to be extracted for further processing (object recognition etc.).

## backjumping
## backmarking
## backoff model
## backpropagation

To minimize the cost function, we need to know how the changes in weights and biases affect the cost function i.e. partial derivatives of the cost function w.r.t every weight and bias in the network; back propagation is a method that allows us to quickly compute all these partial derivatives.

## Backus-Naur form (BNF)

A mathematical notation used to describe the syntax of a programming language.  

## backward-chaining

## bag of words

The **bag of words** model is a simplifying representation used in natural language processing and information retrieval. Also known as the vector space model. In this model, a text is represented as the bag of its words, disregarding grammar and even word order but keeping multiplicity.

## bagging

Bagging (stands for **B**ootstrap **Agg**regat**ing**) is a way to decrease the variance of your prediction by generating additional data for training from your original dataset using combinations with repetitions to produce multisets of the same cardinality/size as your original data. By increasing the size of your training set you can't improve the model predictive force, but just decrease the variance, narrowly tuning the prediction to expected outcome.

## bang-bang control
## baseline
## batch gradient descent
## Bayes' rule

**Bayes' rule** describes the probabilty of an event(lets say A) in the light of that a given event B has already occured.
Mathematically Bayes' rule can be described as :-  **P(A|B) = P(A)P(B|A)/P(B)**

## Bayes-Nash equilibrium
## Bayesian learning

A Machine Learning method which enables us to encode our initial perception of what a model should look like, regardless of what the data tells us. It proves to be very useful when there’s a sparse amount of data to train our model properly.

## Bayesian network

A probabilistic graphical model representing a group of variables along with their conditional dependencies through a direct acyclic graph; it is also used to compute the probability distribution for a subset of network variables, provided the distributions or values of any subset of the remaining variables.

## beam search

Beam search is a heuristic search algorithm that explores a graph by expanding the most promising node in a limited set. Beam search is an optimization of best-first search that reduces its memory requirements. Best-first search is a graph search which orders all partial solutions (states) according to some heuristic. But in beam search, only a predetermined number of best partial solutions are kept as candidates.[1] It is thus a greedy algorithm.

## behaviorism
## belief function
## belief propagation
## belief revision
## belief state
## Bellman equation
## Bellman update

## benchmarking

Benchmarking is to measure the quality of something for the purposes of comparison or evaluation.

## best-first search
## biconditional
## binary constraint
## binary resolution
## binding list
## binocular stereopsis
## biological naturalism
## blocks world
## bluff
## body
## boid

Boids is an artificial life program, developed by Craig Reynolds in 1986, which simulates the flocking behaviour of birds. His paper on this topic was published in 1987 in the proceedings of the ACM SIGGRAPH conference. The name "boid" corresponds to a shortened version of "bird-oid object", which refers to a bird-like object.

## boosting

Boosting is a two-step approach, where one first uses subsets of the original data to produce a series of averagely performing models and then "boosts" their performance by combining them together using a particular cost function (=majority vote). Unlike bagging, in the classical boosting the subset creation is not random and depends upon the performance of the previous models: every new subsets contains the elements that were (likely to be) misclassified by previous models.

## boundary set
## bounded optimality
## bounded PlanSAT
## bounded rationality
## bounds consistent
## bounds propagation
## branching factor
## bridge
## bunch

# C

## calculative rationality
## canonical distribution
## cart-pole

A pole is attached by an un-actuated joint to a cart, which moves along a frictionless track. The system is controlled by applying a force to the cart in the left or right direction. The pendulum starts upright, and the goal is to prevent it from falling over. A reward is provided for every timestep that the pole remains upright.

## cascaded finite-state transducers
## case agreement
## causal
## causal link
## causal network
## center
## central limit theorem

The central limit theorem (CLT) establishes that, in some situations, when independent random variables are added, their properly normalized sum tends toward a normal distribution (informally a "bell curve") even if the original variables themselves are not normally distributed.

## certainty effect
## certainty equivalent
## CFG
## chain rule

It is a mathematical formula used to compute the derivatives of a composition of two or more functions.

## characters

A character is any letter, number, space, punctuation mark, or a symbol.

## chart
## checkers
## chess

Chess is a two-player strategy board game played on a chessboard, a checkered gameboard with 64 squares arranged in an 8×8 grid. Play does not involve hidden information. Each player begins with 16 pieces: one king, one queen, two rooks, two knights, two bishops, and eight pawns. Each of the six piece types moves differently, with the most powerful being the queen and the least powerful the pawn. The objective is to checkmate the opponent's king by placing it under an inescapable threat of capture. To this end, a player's pieces are used to attack and capture the opponent's pieces, while supporting each other.

## Chomsky Normal Form

A grammar is in **Chomsky Normal Form (usually found as CNF)** if all its production rules are in one of the following forms:

```
A -> BC
A -> a
S -> ε
```

where `S` is the starting symbol and `ε` the symbol for the empty string.

## circuit verification
## circumscription
## Clark Normal Form
## classification

Sorting or dividing data into two or more categories on the basis of a distinct feature.

## clause
## closed-loop
## clustering

In terms of Data Science, clustering is the grouping of data instances or objects with similar features and characteristics.   

## clutter
## CMAC
## co-NP
## co-NP-complete
## coarse-to-fine
## coarticulation
## coastal navigation
## cognitive psychology

It is a branch of Psychology which deals with the mental processes involved in obtaining and comprehending new information. Some of the prominent processes include judging, problem solving and remembering.

## collusion
## color constancy
## communication
## commutativity
## competitive
## competitive ratio
## complementary literals
## complete assignment
## complete data
## completeness
## completing the square
## completion
## compliant motion
## composition
## compositional semantics
## compositionality
## computable
## computational linguistics
## computational neuroscience
## conclusion
## concurrent action list
## conditional effect
## conditional Gaussian
## conditional probability table
## conditional random field
## conditioning
## confirmation
## conflict set
## conflict-directed backjumping
## conformant
## conjugate gradient
## conjunct ordering
## conjunction
## conjunctive normal form
## connectionist
## consciousness
## consequentialism
## consistency
## consistent
## consistent plan

A plan in which there are no cycles in the *ordering constraints* and no conflicts with the **causal links**.  

## constraint language
## constraint learning
## constraint logic programming
## constraint optimization problem
## constraint propagation
## constraint satisfaction problem
## constraint weighting
## consumable
## context-free grammar
## context-specific independence
## contingency plan
## continuous
## contraction
## contradiction
## control theory
## controller
## convention
## convex set
## convolution

A mathematical term which basically means merging two signals to form a third signal.

## cooperative
## coordination
## corpus
## Cournot competition
## covariance
## covariance matrix
## critic
## critical path
## critical path  method
## cross-correlation
## crossover point
## cryptarithmetic
## cumulative distribution
## cumulative probability density function
## current-best-hypothesis
## cycle cutset
## cyclic solution
## CYK algorithm

# D
## DARPA Grand Challenge

The DARPA Grand Challenge is a prize competition for American Autonomous Vehicles,funded by the **Defense Advanced Research Projects Agency**,the most prominent research organization of the United States Department of Defense.   

## data association
## data complexity
## data compression

It is the process of encoding data using fewer bits than were used in the original representation so that the data consumes lesser disk space.

## data matrix
## data mining
## data-driven
## database semantics
## Datalog
## Davis-Putnam algorithm
## decayed MCMC
## decentralized planning
## decision analysis
## decision boundary
## decision maker
## decision network
## decision theory
## decision theory
## decision tree

A decision tree is a construct that uses a tree like graph or model of decisions and their possible consequences,including chance event outcomes,resource costs and utility.

## declarative
## declarative bias
## decomposition
## deduction theorem
## deductive learning

Going from a known
general rule to a new rule that is logically entailed (and thus nothing new), but is nevertheless useful
because it allows more efficient processing.  

## deep belief networks
## deep learning

It is a subfield of Machine Learning that tries to map the working of the human brain in processing data and creating patterns to use in decision making.

## default logic
## definite clause
## definite clause grammar
## definition of a rational agent
## deformable template
## degree of belief
## degree of freedom

It is a statistics term which represents the number of variables that you are allowed to change for analysis, without any constraint violation.

## delete list
## deliberative layer
## demonic nondeterminism
## Dempster-Shafer theory
## depth
## depth of field
## depth-first search

It is an algorithm which allows us to traverse a graph or a tree data structure; it starts from the root node and traverses as far as possible for each branch before backtracking. (In case of graph data structure, the root node would be any arbitrary node that you select).

## depth-limited search
## detailed balance
## Deterministic
## diachronic
## diagnostic
## diameter
## Differential GPS
## diffuse albedo
## Diophantine equations
## direct utility estimation
## Dirichlet process
## disambiguation
## discount factor
## discrete
## discretization
## disjoint
## disjunction
## disjunctive constraint
## disparity
## distant point light source
## distortion
## distributed constraint satisfaction
## DL
## domain
## domain closure
## dominant strategy equilibrium
## downward refinement property
## dropping conditions
## DT
## dual graph
## dualism
## duration
## dynamic
## dynamic backtracking
## dynamic Bayesian network
## dynamic programming

A method of solving complex problems by breaking them down to sub-problems that can be solved by back tracking from the last stage. Used in popular real-world problems including traveling salesman problem, Fibonacci sequence, knapsack problem, etc.

## dynamic state

# E

## early stopping
## economy
## effect
## effective branching factor
## efficient
## electric motor
## eliminative materialism
## elitism
## embodied cognition
## emergent behavior
## empirical gradient
## empirical loss
## empiricism
## English auction
## entailment
## entropy
## environment
## environment generator
## episodic
## epsilon-ball
## equality symbol
## equilibrium
## ergodic
## error rate
## event
## evidence
## evidence reversal
## evolutionary algorithms
## evolutionary psychology
## evolutionary strategies
## exact cell decomposition
## execution
## execution monitoring
## executive layer
## exhaustive decomposition
## existence uncertainty
## Existential Instantiation
## expand
## expectation
## expectation-maximization
## expected value
## expectiminimax value
## explanation-based learning
## explanatory gap
## exploitation
## exploration
## exploration problem
## expressiveness
## extended Kalman filter (EKF)
## extension
## extensive form
## externalities
## extrinsic

# F
## fact
## factor
## factored frontier
## factored representation
## factorial HMM
## factoring
## false negative
## false positive
## feature extraction
## feature selection
## feed-forward network
## FIFO queue
## filtering
## finite horizon
## first-choice hill climbing
## first-order Markov process
## fixate
## fixed point
## fixed-lag smoothing
## flaw
## fluent
## focal plane
## foreshortening
## forward-backward algorithm
## forward-chaining
## frame problem
## frames
## framing effect
## free space
## frequentist
## friendly AI
## frontier
## full joint probability distribution
## fully observable

If an agent's sensors give it access to the complete state of the environment at each point in time,then we say that the task environment is fully observable.

## functionalism
## futility pruning
## fuzzy control
A fuzzy control system is a control system based on fuzzy logic—a mathematical system that analyzes analog input values in terms of logical variables that take on continuous values between 0 and 1, in contrast to classical or digital logic, which operates on discrete values of either 1 or 0 (true or false, respectively).

## fuzzy logic

Fuzzy logic is a form of many-valued logic in which the truth values of variables may be any real number between 0 and 1. It is employed to handle the concept of partial truth, where the truth value may range between completely true and completely false.

## fuzzy set theory

Fuzzy sets (aka uncertain sets) are somewhat like sets whose elements have degrees of membership. In classical set theory, the membership of elements in a set is assessed in binary terms according to a bivalent condition — an element either belongs or does not belong to the set. By contrast, fuzzy set theory permits the gradual assessment of the membership of elements in a set; this is described with the aid of a membership function valued in the real unit interval [0, 1].

# G

## G-set
## gain parameter
## gain ratio
## gait
## game theory

Game theory is the study of mathematical models of strategic interaction between rational decision-makers. It has applications in all fields of social science, as well as in logic and computer science. Originally, it addressed zero-sum games, in which one person's gains result in losses for the other participants. Today, game theory applies to a wide range of behavioral relations, and is now an umbrella term for the science of logical decision making in humans, animals, and computers.

## game tree
## Gaussian distribution
## Gaussian error model
## Gaussian filter
## Gaussian process
## generalization
## generalization hierarchy
## generalization loss
## generalized modus ponens
## generating
## generator
## genetic algorithms
## genetic programming
## Gibbs sampling
## GLIE
## global constraint
## global minimum
## Go
## goal
## goal clauses
## goal formulation
## goal monitoring
## goal test
## goal-directed reasoning
## gold standard
## gorilla problem
## gradient
## gradient descent
## grammar
## graph
## graph coloring
## grasping
## greedy agent
## greedy best-first search
## grid world
## ground term
## grounding

# H

## Hamming distance
## Hansard
## haptic feedback
## head
## heavy-tailed distribution
## Hebbian learning
## Hessian
## heuristic function
## heuristic search
## hidden Markov model

A hidden Markov model (or HMM) is a temporal probabilistic model in which the state of the process is described by a *single discrete*
random variable.

## hierarchical lookahead
## hierarchical reinforcement learning
## high-level action
## Hinton diagrams
## holdout cross-validation
## homeostatic
## homophones
## horizon effect
## Horn clause
## hub
## human-level AI
## Hungarian algorithm
## hybrid A*
## hybrid agent
## hybrid architecture
## hybrid Bayesian network
## hydraulic actuation
## hypothesis
## hypothesis prior
## hypothesis space

# I

## i.i.d.

**i.i.d.** denotes **independent and identically distributed** random variables. They are defined on the same probability space, have identical probability distribution functions, and are mutually independent.

## identification in the limit
## identity matrix
## identity uncertainty
## ignore delete lists
## ignore preconditions heuristic
## image
## imperfect information
## implementation
## implementation level
## implication
## importance sampling
## inclusion-exclusion principle
## incompleteness theorem
## incremental belief-state search
## independence
## independent subproblems
## index
## indexed random variable
## indexing
## individuation
## induction
## inductive learning

Going from a set of specific input-output pairs
to a (possibly incorrect) general rule  is called **inductive learning**.

## inductive logic
## inductive logic programming
## inference
## inference rules
## inferential frame problem
## infinite
## infinite horizon
## infix
## information extraction
## information gain
## information gathering
## information retrieval
## information sets
## informed search
## inheritance
## initial state
## input resolution
## inside-outside algorithm
## insurance premium
## intelligence
## interleaving
## interlingua
## internal state
## interpretation
## interreflections
## intrinsic
## intuition pump
## inverse
## inverse entailment
## inverse kinematics
## inverse reinforcement learning
## inverted pendulum
## inverted spectrum
## IR
## irreversible
## iterative deepening search
## iterative expansion
## iterative-deepening A*

# J

## join tree
## joint action
## joint plan
## JTMS
## justification

# K

## k-d tree
## K-means clustering
## Kalman filtering
## Kalman gain matrix
## kernel
## kernel function
## kernel trick
## kinematic state
## kinematics
## King Midas problem
## knowledge acquisition
## knowledge base
## knowledge engineering
## knowledge-based agents
## Known
## Kriegspiel
## Kullback-Leibler divergence

# L

## label
## Lambert's cosine law
## landmarks
## language
## language generation
## language identification
## large-scale learning
## layers
## leak node
## learning

An agent is **learning** if it improves its
performance after making observations about the world.

## learning curve
## learning element
## learning rate
## least commitment
## least-constraining-value
## leave-one-out cross-validation
## lens
## level cost
## level of abstraction
## level sum
## leveled off
## lexical category
## lexicon
## LIFO queue
## lifting lemma
## likelihood
## likelihood weighting
## line search
## linear Gaussian
## linear interpolation smoothing
## linear programming
## linear regression
## linear resolution
## linear separator
## linkage constraints
## links
## liquid event
## Lisp
## literal
## local consistency
## local search
## locality
## locality-sensitive hash
## localization
## locally structured
## locally weighted regression
## location sensors
## locking
## log likelihood
## logic
## logical equivalence
## logical minimization
## logical omniscience
## logicist
## logistic function
## logistic regression
## long-distance dependencies
## LOOCV
## loopy path
## loosely coupled
## loss function
## lottery
## low-dimensional embedding

# M

## machine reading
## macrops
## magic set
## Mahalanobis distance
## Maintaining Arc Consistency (MAC)
## makespan
## margin
## marginalization
## Markov blanket
## Markov chain
## Markov decision process
## Markov localization
## Markov network
## Markov property
## material value
## materialism
## matrix
## max norm
## max-level
## maximin
## maximin equilibrium
## maximum a posteriori
## maximum expected utility
## maximum-likelihood
## mechanism
## mechanism design
## mel frequency cepstral coefficient (MFCC)
## memoization
## memoized
## mental states
## mereology
## metadata
## metalevel learning
## metalevel state space
## metaphor
## metareasoning
## metonymy
## micromort
## min-conflicts
## mind-body problem
## minimax
## minimax decision
**minimax decision** is the optimal choice which leads MAX to the state with the highest minimax value and leads MIN to lowest minimax value.
## minimax search
## minimax value
The **minimax value** of a node in a game tree is the utility (for MAX) of being in the corresponding state, assuming that both players play optimally from there to the end of the game.
## minimum description length
## minimum slack
## minimum-remaining-values
## Minkowski distance
## missing precondition
## missing state variable
## mixture distribution
## mixture of Gaussians
## mobile manipulator
## modal logic
## model
## model checking
## model selection
## modus ponens
## monitoring
## monotonic preference
## monotonicity
## Monte Carlo
## Monte Carlo localization
## Monte Carlo simulation
## Monte Carlo tree search
## motion blur
## motion model
## multiactor
## multiagent
## multiagent planning problem
## multiagent systems
## multiplexer
## multiplicative utility function
## multiply connected
## multivariate Gaussian
## multivariate linear regression
## mutation
## mutex
## mutual preferential independence
## mutually utility independent
## myopic

# N

## n-armed bandit
## n-gram model
## natural kind
## natural numbers
## nearest-neighbor filter

The nearest-neighbour filter, which repeatedly chooses the closest pairing of predicted position and observation and adds that pairing to the assignment.

## nearest-neighbors regression
## negation
## negative
## neuroscience
## Newton-Raphson
## no-good
## no-regret learning
## noise
## noisy channel model
## noisy-OR
## nondeterministic
## nonholonomic
## nonlinear
## nonlinear regression
## nonmonotonicity
## nonparametric
## nonparametric density estimation
## nonparametric model
## normalization
## normalized form
## normative theory
## NP-complete
## NP-completeness
## null hypothesis

# O

## object model
## objective function
## objectivist
## occupancy grid
## occupied space
## occur check
## Ockham's razor
## odometry
## off-policy
## omniscience
## on-policy
## online replanning
## online search
## ontological commitment
## ontological engineering
## ontology
## open list
## open-code
## open-loop
## operationality
## operations research
## optimal brain damage
## optimal controllers
## optimally efficient
## optimization
## optimizer's curse
## optogenetics
## ordering constraints
## OR-parallelism
## orientation
## origin function
## Othello
## out of vocabulary
## outcome
## overall intensity
## overfitting

# P

## PAC learning
## PageRank
## parameter independence
## parameter learning
## parametric model
## Pareto dominated
## parse tree
## parsing
## partial assignment
## partial information
## partial program
## partially observable
## particle filtering
## partition
## passive learning

A passive learning agent learns from its observations, but the actions the agent takes are not influenced by the learning process.
This is in contrast to an active learning agent, which chooses actions that will facilitate its own learning.

## path
## path planning
## paths
## pattern matching
## payoff function
## PD controller
## PDDL
## Peano axioms
## PEAS
## peeking
## percept

The term **percept** refers to the agent's perceptual inputs at any given instant.

## percept schema
## percept sequence

An agent's **percept sequence** is the complete history of everything the agent has ever perceived.

## perception
## perception layer
## perceptron
## perceptron network
## perfect rationality
## performance element
## perplexity
## persistence arc
## persistent failure model
## perspective projection
## phone model
## phoneme
## phrase structure
## physical symbol system
## physicalism
## piano movers
## pictorial structure model
## PID controller
## plan monitoring
## plan recognition
## planning graph
## PlanSAT
## playout
## ply
## pneumatic actuation
## point-to-point motion
## poker
## policy
## policy evaluation
## policy gradient
## policy improvement
## policy iteration
## policy loss
## policy search
## policy value
## polynomial kernel
## pose
## positive
## possibility axiom
## possibility theory
## possible world
## post-decision disappointment
## pragmatics
## precedence constraints
## precision

**Precision** is a performance measure often used to describe some model, alongside other measures like *accuracy*, *recall* etc. Precision can be thought of as an efficiency measure of a model. It is given as:  
***Precision = True Positives / (True Positives + False Positives)***

## precondition
## prediction
## preference elicitation
## preference independence
## prefix
## premise
## presentation
## principle of indifference
## principle of insufficient reason
## principle of trichromacy
## prioritized sweeping
## priority queue
## prisoner's dilemma
## probabilistic checkmate
## probabilistic Horn abduction
## probabilistic inference
## probability
## probability density function
## probability distribution
## probability model
## probit distribution
## problem
## problem formulation
## problem-solving agent
## procedural attachment
## process
## product rule
## progression planning
## Prolog
## pronunciation model
## proof
## proof-checker
## proposition symbol
## propositionalize
## protein design
## provably beneficial
## pruning
## psychological reasoning
## PUMA
## pure strategy
## pure symbol

# Q

## Q-learning
## QALY
## quadratic programming
## qualia
## qualification problem
## qualitative physics
## quantification
## quantization factor
## quasi-logical form
## question answering
## queue
## quiescence search

# R

## radial basis function
## radiometry
## random surfer model
## random-restart hill climbing
## randomized weighted majority algorithm
## rational agent

A rational agent selects an action that is expected to maximize its performance measure,given the evidence provided by the
*percept sequence* and whatever built-in knowledge the agent has.

## rationalism
## rationality
## reachable set
## reactive control
## reactive layer
## real-time AI
## realizable
## reasoning
## recall

**Recall** is a measure of performance used alongside ***Precision***, ***Accuracy*** and ***F-score***. It is defined as the ratio of the *true positives* to the *summation of true positives and false negatives*.

## reciprocal rank
## recognition
## recombine
## reconstruction
## record linkage
## rectangular grid
## recurrent network
## recursive
## recursive best-first search
## reduct
## reference class
## reference controller
## reference path
## reflect
## reflective architecture
## refutation
## regions
## regression
## regression planning
## regression to the mean
## regret
## regular expression
## regularization
## reinforcement
## reinforcement learning

In **reinforcement learning** the agent learns from a series of
reinforcements-rewards or punishments.

## rejection sampling
## relational extraction
## relational uncertainty
## relative error
## relative likelihood
## relaxed problem
## relevance
## relevance feedback
## relevant
## relevant-states
## renaming
## rendering
## rendering model
## repeated state
## resolution
## resolvent
## result set
## Rete algorithm
## retrograde
## reusable
## revelation principle
## revenue equivalence theorem
## reward
## reward shaping
## reward-to-go
## risk-averse
## risk-neutral
## risk-seeking
## Robocup
## robot navigation
## robotic soccer
## robust control theory
## ROC curve
## rollout
## Roomba
## Root Mean Square (RMS)

A mathematical formula, **root mean square** is often used as an error estimator. It is described as the root of the summation of all the squared errors. The formula is given as:
<br>***square_root[summation(a1^2 + a2^2 + ...)], where a1, a2...are some entities***
<br> In RMS error estimation, the above squared entities are replaced with squared errors.


## rules

# S

## S-set
## sample complexity
## sample space
## sampling rate
## SARSA
## SAT
## satisfiability
## satisfiability threshold conjecture
## satisficing
## scaled orthographic projection
## scanning lidars
## scene
## schedule
## schedulers
## schema
## Scrabble
## sealed-bid second-price auction
## search
## search cost
## search tree
## segmentation
## selection
## semantic ambiguity
## semantics
## semi-supervised learning
## semidynamic
## semiotics
## sensitivity analysis
## sensor interface layer
## sensor Markov assumption
## sensorless
## sequence form
## sequential
## sequential Monte Carlo
## set of support
## set semantics
## set-cover problem
## set-level
## shading
## shadow
## shape
## shaving
## shortcuts
## shoulder
## sibyl attack
## sideways move
## sigmoid perceptron
## significance test
## similarity networks
## simulated annealing
## simultaneous localization and mapping (SLAM)
## single agent
## singly connected
## singular
## singularity
## situation
## situation calculus
## skeletonization
## Skolemization

**Skolemization** is the process of removing the existential quantifiers by elimination. This is similar to the inference rule ***Existential Elimination*** where an inference involving sentence *a*, variable *v* and constant *k* can be made provided *k* does not occur anywhere in the knowledge base.

## slack
## slant
## sliding window
## small-scale learning
## smoothing

**Smoothing** is the process of computing the distribution over past states given evidence up to the present.

## soccer
## social laws
## Socratic reasoner
## soft margin
## softmax function

The **softmax function** is a mathematical function often used for classification tasks. This function calculates the probability distribution of one class, over all the available classes. It's formula is given as:

*F(X) = exp(X) / summation(exp(X))* with the summation being over all the classes.

## software architecture
## sokoban
## solution
## sonar sensors
## sound
## spam detection
## sparse
## sparse model
## spatial reasoning
## specialization
## specular reflection
## specularities
## speech act
## Speech Recognition

**Speech recognition** is the process of analyzing audio and ***recognizing*** parts related to speech within the audio file. This process of ***recognition*** may involve simply identifying the speech part, gender identification and also as complex as identifying the words spoken when given an audio. This field often overlaps into the domain of artificial intelligence and machine learning.

## split point
## stable
## stable model
## standard normal distribution
## standardizing apart
## Starcraft
## start symbol
## state abstraction
## state estimation
## state space
## state-space landscape
## static
## stationarity assumption
## stationary distribution
## stationary process
## stemming
## step cost
## step size
## stereo vision
## stochastic
## stochastic beam search
## stochastic games
## stochastic hill climbing
## stochastic policy
## straight-line distance
## strategic form
## strategy
## strategy profile
## strategy-proof
## strong AI
## structural EM
## structured representation
## stuff
## subcategory
## subgoal independence
## subject-verb agreement
## subjectivist
## subproblem
## substitution
## subsumption
## subsumption architecture
## subsumption lattice
## successor
## successor-state axiom
## Sudoku
## sum of squared differences
## superpixels
## supervised learning

In **supervised learning** the agent observes some example
input-output pairs and learns a function that maps from input to
output.

## support vector machine
## symmetry-breaking constraint
## synchro drive
## synchronic
## synchronization
## syntactic ambiguity
## syntactic theory
## syntax
## synthesis

# T

## table lookup
## tabu search
## tactile sensors
## taxonomy
## Taylor expansion
## technological singularity
## template
## temporal logic
## temporal-difference
## temporal-projection
## term
## terminal states
## terminal test
## test set
## text classification
## texture
## theorem proving
## thrashing
## tiling
## tilt
## time and tense
## time line
## time of flight camera
## time to answer
## tit-for-tat
## topological sort
## total Turing Test
## toy problem
## trace
## tractability
## tragedy of the commons
## trail
## training curve
## training set
## transfer model
## transhumanism
## transition model
## transition probability
## transpose
## transposition table
## traveling salesperson problem
## tree decomposition
## tree width
## treebank
## truth
## truth value
## truth-preserving
## truth-revealing
## turbo decoding
## Turing Test
  The **Turing Test** is a test proposed by Alan Turing in 1950, which is used to determine whether a computer is intelligent by evaluating the "human-ness" of its responses.
## type A strategy
## type B strategy
## type constraint
## type signature

# U

## ultraintelligent machine
## unary constraint
## unbiased
## uncertainty
## underconstrained
## understanding
## unification
## unifier
## uniform-cost search
## Unimate
## uninformed search
## unique action axioms
## unique string axiom
## unit clause
## unit preference
## unit propagation
## unit resolution
## units function
## universal grammar
## Universal Instantiation
## unknown
## unobservable
## unrolling
## unsupervised clustering
## unsupervised learning

In **unsupervised learning**
the agent learns patterns in the input without any explicit feedback.

## upper confidence bounds on trees
## upper ontology
## Urban Challenge
## utility
## utility independence

# V

## vague
## validation set

A part of the dataset that is used to tune the parameters of a machine learning model. It can also be used to determine a stopping point for the back-propagation algorithm.

## validity
## value
## value alignment
## vanishing point

A vanishing point of a function means the function has a zero at the point or on the set.

## variable

A variable is a symbol on whose value a function, polynomial, etc., depends.

## variational approximation
## variational parameters
## VCG
## vector

A vector is formally defined as an element of a vector space R^n. A vector is given by n coordinates and can be specified as (A_1,A_2,...,A_n).

## vector field histograms
## vehicle interface layer
## verification
## version space
## Vickrey-Clarke-Groves
## virtual counts
## virtual support vector machine
## VLSI layout
## vocabulary
## Voronoi graph

# W

## weak AI
## weak learning
## weight
## weight space
## weighted A* search
## weighted training set
## wide content
## Widrow-Hoff rule
## Winnow algorithm
## workspace representation
## wrapper
## wumpus world

# Z

## zero-sum games

In game theory and economic theory, a **zero-sum game** is a mathematical representation of a situation in which each participant's gain or loss of utility is exactly balanced by the losses or gains of the utility of the other participants. If the total gains of the participants are added up and the total losses are subtracted, they will sum to zero.