Abstract: One aspect of the invention is the construction of mixtures of Bayesian networks. Another aspect of the invention is the use of such mixtures of Bayesian networks to perform inferencing. A mixture of Bayesian networks (MBN) consists of plural hypothesis-specific Bayesian networks (HSBNs) having possibly hidden and observed variables. A common external hidden variable is associated with the MBN, but is not included in any of the HSBNs. The number of HSBNs in the MBN corresponds to the number of states of the common external hidden variable, and each HSBN is based upon the hypothesis that the common external hidden variable is in a corresponding one of those states. In one mode of the invention, the MBN having the highest MBN score is selected for use in performing inferencing.

Abstract: Distribution displays for categories are provided which illuminate the distribution of continuous attributes over all cases in a category, and which provide a histogram of the population of the different states of categorical attributes. An array of such displays by attribute (in one dimension) and category (in another dimension) may be provided. Category diagram displays are also provided for visualizing the different categories, and their distributions, populations, and similarities. These are displayed through different shading of nodes and edges representing categories and the relationship between two categories, and through proximity of nodes.

Abstract: Systems and methods are provided for producing displays of the accuracy of data mining or statistical models that produce associative predictions. For all cases in a testing data set, the model makes predictions and provides associated probabilities. The cases are sorted by their probability of making accurate predictions and a graph is made of the accuracy of the model over various subsets containing the highest probability cases as evaluated by the model. Where a number of probabilities are presented for the predictions in a basket of predictions, those probabilities are combined to yield a probability score for the entire basket. Additionally, the accuracy of a model over different basket sizes may be graphed. The accuracy graph may also be produced for any models making a prediction, by graphing the probability of making accurate predictions and a graph made of the accuracy of the model over various subsets of the data containing the highest probability cases.

Abstract: The present invention includes a system and a method for processing large data sets that are too large to conveniently fit into a formal database application. The large data set processing system and method use a prediction model having a feature selection capability to process a fraction of the large data set and define useful predictors. The useful features are used to make predictions for the entire data set. The large data set processing system includes a useful predictor module, for defining useful predictors, and a feature-selection prediction model, for processing a portion of the data set (including the useful predictors) to obtain prediction results.

Abstract: The invention automatically determines the number of clusters in a Bayesian network or in a mixture of Bayesian networks (MBN). A common external hidden variable is associated with the network. Expected sufficient statistics (ESS) are computed in the case of a Bayesian network or expected complete model sufficient statistics (ECMSS) are computed in the case of an MBN, from the observed data. An expected sample size for each state of a hidden variable is computed from the ESS or ECMSS. The optimum number of states is reached by deleting those states having a sample size less than a predetermined threshold.

Abstract: One aspect of the invention is the construction of mixtures of Bayesian networks. Another aspect of the invention is the use of such mixtures of Bayesian networks to perform inferencing. A mixture of Bayesian networks (MBN) consists of plural hypothesis-specific Bayesian networks (HSBNs) having possibly hidden and observed variables. A common external hidden variable is associated with the MBN, but is not included in any of the HSBNs. The number of HSBNs in the MBN corresponds to the number of states of the common external hidden variable, and each HSBN is based upon the hypothesis that the common external hidden variable is in a corresponding one of those states. In one mode of the invention, the MBN having the highest MBN score is selected for use in performing inferencing.

Abstract: One aspect of the invention is the construction of mixtures of Bayesian networks. Another aspect of the invention is the use of such mixtures of Bayesian networks to perform inferencing. A mixture of Bayesian networks (MBN) consists of plural hypothesis-specific Bayesian networks (HSBNs) having possibly hidden and observed variables. A common external hidden variable is associated with the MBN, but is not included in any of the HSBNs. The number of HSBNs in the MBN corresponds to the number of states of the common external hidden variable, and each HSBN is based upon the hypothesis that the common external hidden variable is in a corresponding one of those states. In one mode of the invention, the MBN having the highest MBN score is selected for use in performing inferencing.

Abstract: The invention employs mixtures of Bayesian networks to perform clustering. A mixture of Bayesian networks (MBN) consists of plural hypothesis-specific Bayesian networks (HSBNs) having possibly hidden and observed variables. A common external hidden variable is associated with the MBN, but is not included in any of the HSBNs. The number of HSBNs in the MBN corresponds to the number of states of the common external hidden variable, and each HSBN is based upon the hypothesis that the common external hidden variable is in a corresponding one of those states. In one mode of the invention, the MBN having the highest MBN score is selected for use in performing inferencing. The invention determines membership of an individual case in a cluster based upon a set of data of plural individual cases by first learning the structure and parameters of an MBN given that data and then using the MBN to compute the probability of each HSBN generating the data of the individual case.

Abstract: The invention performs speech recognition using an array of mixtures of Bayesian networks. A mixture of Bayesian networks (MBN) consists of plural hypothesis-specific Bayesian networks (HSBNs) having possibly hidden and observed variables. A common external hidden variable is associated with the MBN, but is not included in any of the HSBNs. The number of HSBNs in the MBN corresponds to the number of states of the common external hidden variable, and each HSBN models the world under the hypothesis that the common external hidden variable is in a corresponding one of those states. In accordance with the invention, the MBNs encode the probabilities of observing the sets of acoustic observations given the utterance of a respective one of said parts of speech. Each of the HSBNs encodes the probabilities of observing the sets of acoustic observations given the utterance of a respective one of the parts of speech and given a hidden common variable being in a particular state.