AI 용어집
인공지능 완전 사전
Dynamic Bayesian Networks
Temporal extension of static Bayesian networks modeling the evolution of random variables and their conditional dependencies over time through Markovian transitions.
Hidden Markov Chains
Statistical models where the observed system depends on non-observable hidden states following a Markov chain, used for time sequence recognition.
Factorial Hidden Markov Models
Extension of HMM where multiple hidden Markov chains interact to generate observations, allowing modeling of complex dependencies between latent factors.
Particle Filter
Sequential Monte Carlo method for state estimation in non-linear and non-Gaussian systems, using a set of weighted particles to approximate the posterior distribution.
Dynamic Topic Models
Extension of topic models that capture the temporal evolution of themes in textual corpora, modeling how word distributions change over time.
Hybrid Dynamic Systems
Models combining discrete and continuous variables in a temporal framework, capturing interactions between discrete states and continuous evolutions in complex systems.
State Space Models
General framework for modeling time series with unobserved latent states evolving according to a state equation and generating observations via an observation equation.
Forward/Backward Inference
Algorithms for computing belief distributions in dynamic graphical models, combining forward propagation (filtering) and backward (smoothing) for optimal state estimation.
Regime Change Models
Models where system parameters can change abruptly or gradually between different behavioral regimes, capturing structural transitions in data.
Change Detection Process
Statistical algorithms to identify time points where statistical properties of a signal or system change significantly, essential for real-time monitoring.
Hierarchical Temporal Graph Models
Multi-scale structures capturing temporal dependencies at different abstraction levels, enabling efficient modeling of complex phenomena with nested dynamics.
Bayesian Filters
General theoretical framework for recursive estimation of a dynamic system state using Bayes' theorem, including Kalman filters and particle filters as special cases.
Temporal Transition Models
Components of dynamic graphical models specifying how probability distributions of variables evolve between successive time steps, defining the system dynamics.
Temporal Influence Networks
Probabilistic models capturing how causal influences between variables propagate and modify over time, essential for analysis of dynamic causal systems.
Non-Stationary Graph Models
Extensions of dynamic graphical models where parameters and/or graph structure can change over time, adapting the model to non-homogeneous evolutions.
Dynamic Structure Learning
Process of automatic inference of optimal structure of a dynamic graphical model from temporal data, including discovery of dependencies and temporal delays.
Viterbi Smoothing
Algorithm to find the most probable sequence of hidden states given all observations, generalizing the Viterbi algorithm for offline inference in dynamic models.
Hidden Markov Models with Inputs
Extension of HMMs where state transitions depend on observed external input variables, allowing conditional modeling of temporal sequences.