|
|
|
|
|
|
|
|
|
Harpy
| Глоссарий по искусственному интеллекту |
B. lowerre wrote harpy, a speech understanding system, in 1975. this program contained a series of precompiled words, and the "understanding" was implemented as a series of transitions between these words. hart-j and hart-s two modular hierarchial art (hart) models. the first implements an agglomerative clustering method, while the second implements a divisive clostering method, where each art layer learns the differences between the input and the matching prototype of the previous layer. see also: ftp:://ftp.sas.com/pub/neural/faq2.html, http://www.wi.leidenuniv.nl/art/.
|
|
Understanding, английский
- Интерпретация
- N когн. понимание (син. comprehension) ungrammatical a неправильно построенное (предложение ) (син. ill formed)
Implements, английский
Инструменты; орудия, инвентарь, принадлежности
Clustering, английский
- The tendency e.g. of laser-active ions in laser gain media to form clusters in their host medium
- Кластеризация; выделение кластеров; объединение в кластеры; группирование
- Clustering is a technique used in data mining, online analytical processing (olap), and similar activities to group observations. a clustering procedure attempts to identify and extract "similar" groups of observations from a data set. unlike classification procedures, the number and memberships in the groups are not known a priori. the observations are grouped on the same set of attributes as is used to assess their homogeneity. in contrast, classification and regression techniques use one group of variables to obtain homogeneity on a second set. the clustering process can either be agglomerative, where each record starts as its own cluster and is merged with other clusters, or divisive, where all the data starts in a single large cluster and is gradually broken down into smaller, more homogeneous clusters. while the clustering is often of interest in its own right, it can also be used to pick out interesting subgroups for further analysis. the clusters can be disjoint, so that each observation falls in one and only one cluster, hierarchical, where the higher level clusters are further split into sub-clusters recursively, or overlapping, so that an individual data point can have membership into more than one cluster. see also: classification procedures, data mining, knowledge discovery in databases, regression.
- A data mining technique that analyzes data to group records together according to their location within the multidimensional attribute space.
- The grouping of multiple servers in a way that allows them to appear to be a single unit to client computers on a network. clustering is a means of increasing network capacity, providing live backup in case one of the servers fails, and improving data security.
Differences, английский
Цифровой избирательный вызов, цив - технология радиосвязи, при которой предварительное согласование канала вызова другого судна осуществляется не широковещательно на голосовом канале, а на отдельном цифровом канале. после согласования рабочего канал
Understand, английский
|
Autoepistemic logic, английский
Autoepistemic logic is a form of nonmonotone logic developed in the 1980s. it extends first-order logic by adding a new operator that stands for "i know" or "i believe" something. this extension allows introspection, so that if the system knows some fact a, it also knows that it knows a and allows the system to revise its beliefs when it receives new information. variants of autoepistemic logic can also include default logic within the autoepistemic logic. see also: default logic, nonmonotone logic.
Bayesian hierarchical model, английский
Bayesian hierarchical models specify layers of uncertainty on the phenomena being modeled and allow for multi-level heterogeneity in models for attributes. a base model is specified for the lowest level observations, and its parameters are specified by prior distributions for the parameters. each level above this also has a model that can include other parameters or prior distributions.
|
|
|
|
|
|
|
