Glosarium AI
Kamus lengkap Kecerdasan Buatan
Probability factor
Non-negative function that represents a local interaction between variables in a factorized probabilistic model. Factors allow decomposition of a complex joint distribution into products of simpler terms.
Distribution factorization
Mathematical process decomposing a joint probability distribution into a product of simpler local factors. This factorization exploits conditional independencies to reduce computational complexity.
Mixture model
Probabilistic model representing a distribution as a convex combination of multiple component distributions. Each component models a subgroup or cluster in the data with its own parameters.
Principal Component Analysis
Linear dimensionality reduction technique transforming correlated variables into orthogonal uncorrelated components. PCA maximizes explained variance with a minimal number of dimensions.
Non-negative Matrix Factorization
Matrix decomposition algorithm constraining factors to be non-negative, ensuring additive interpretability. Used notably in text analysis and image processing for parts-based representations.
Latent Dirichlet Allocation
Probabilistic generative model for topic discovery in document collections. LDA assumes each document is a mixture of topics and each topic a distribution over words.
Markov Random Field
Undirected graphical model where dependencies are specified by factors on variable cliques. MRFs are particularly suited for output structuring problems and spatial modeling.
Singular Value Decomposition
Matrix factorization decomposing a matrix into a product of three orthogonal and diagonal matrices. SVD provides the best low-rank approximation in the least squares sense and reveals latent structure.
Chow-Liu Factorization
Algorithm that constructs the optimal dependency tree approximating a multivariate distribution. The method maximizes likelihood under tree structure constraint using mutual information weights.