BNL: A Matlab toolbox for Bayesian networks with logistic regression nodes

The BNL toolbox is a set of Matlab functions for defining and estimating the parameters of a Bayesia

Description

The BNL toolbox is a set of Matlab functions for defining and estimating the parameters of a Bayesian network for discrete variables in which the conditional probability tables are specified by logistic regression models. Logistic regression can be used to incorporate restrictions on the conditional probabilities and to account for the effect of covariates. Nominal variables are modeled with multinomial logistic regression, whereas the category probabilities of ordered variables are modeled through a cumulative or adjacent-categories response function. Variables can be observed, partially observed, or hidden.
Additional features include the capability of merging a set of terminal observed nodes that share the same parents into a single node to speed up computations; the possibility of restricting parameters to be equal or to a particular value; and the possibility of incorporating continuous normally distributed latent variables as parents of discrete variables.
Parameters are estimated by an EM-algorithm where the E-step is carried efficiently by operating on a junction tree associated with the Bayesian network. To construct a junction tree from a user-specified directed acyclic graph and to specify a schedule of flows operating on the junction tree during the E-step, BNL calls for the Bayes Net Toolbox (BNT).
BNL can be used to estimate a wide variety of item response, latent class, and latent transition models. The scope of these models is widened considerably by making use of the efficient EM-algorithm.

• MATLAB release: MATLAB 7.2 (R2006a)

Other Files

gausskwad/herzo.m,
generate_data/generate_bnet_data.m,
messagepassing/franksnormalize.m,
messagepassing/franks_init_pot.m,
messagepassing/franks_marginalize_pot.m,
messagepassing/franks_max_marginalize_pot.m,
messagepassing/max_propagate_messages.m,
messagepassing/pot_division.m,
messagepassing/propagate_messages.m,
messagepassing/replicate_pot.m,
messagepassing/select_max_config.m,
BNL manual.pdf,
additional/adj_logistic.m,
additional/adj_logit.m,
additional/cum_logistic.m,
additional/cum_logit.m,
additional/deriv_adj_logist.m,
additional/deriv_cum_logist.m,
additional/deriv_multinom_logist.m,
additional/dummycode.m,
additional/infocrit.m,
additional/multinom_logistic.m,
additional/multinom_logit.m,
additional/multinornd.m,
additional/randvector.m,
constructbnt/franks_from_BNT.m,
constructbnt/franks_mk_adj_mat.m,
constructbnt/inv_order.m,
constructbnt/link_pot_to_CPT.m,
designmatrices/check_order.m,
designmatrices/construct_design_mats.m,
designmatrices/construct_lin_pred.m,
designmatrices/construct_predmat.m,
designmatrices/cov_into_design.m,
designmatrices/define_lin_pred_struct_cov_default.m,
designmatrices/define_lin_pred_struct_cov_main.m,
designmatrices/define_lin_pred_struct_main.asv,
designmatrices/define_lin_pred_struct_main.m,
designmatrices/define_lin_pred_struct_sat.m,
estimation/compute_JPTs.m,
estimation/compute_suff_stats.m,
estimation/compute_suff_stats_ind.m,
estimation/construct_bigCPTs.m,
estimation/construct_equiv_class_CPT.m,
estimation/construct_sCPT.m,
estimation/EM_iteration.m,
estimation/find_max_configs.asv,
estimation/find_max_configs.m,
estimation/fit_multinom_logistic.m,
estimation/fit_ordered_logistic.m,
estimation/gen_random_start.m,
estimation/loglik.m,
estimation/max_marginalization.m,
estimation/num_infomatrix_anal_score.m,
estimation/score.m,
estimation/update_parms.m,
example_models/alarm with restrictions/comparemodels.m,
example_models/alarm with restrictions/construct_alarm.m,
example_models/alarm with restrictions/fit_model_cumul.m,
example_models/alarm with restrictions/fit_model_cumul50.asv,
example_models/alarm with restrictions/fit_model_cumul50.m,
example_models/alarm with restrictions/fit_model_cumul50_test.asv,
example_models/alarm with restrictions/fit_model_cumul50_test.m,
example_models/alarm with restrictions/fit_model_cumul_test.m,
example_models/alarm with restrictions/fit_model_norest.asv,
example_models/alarm with restrictions/fit_model_norest.m,
example_models/alarm with restrictions/fit_model_norest_test.asv,
example_models/alarm with restrictions/fit_model_norest_test.m,
example_models/alarm with restrictions/gen_alarm_start.asv,
example_models/alarm with restrictions/gen_alarm_start.m,
example_models/alarm with restrictions/simulate50_50.m,
example_models/anorex/construct_bnet_hier_hmm.m,
example_models/anorex/construct_bnet_hmm.m,
example_models/anorex/construct_equiv_hier_hmm.m,
example_models/anorex/define_lin_pred_struct_hier_hmm_main.m,
example_models/anorex/equiv_classes_hier_hmm.m,
example_models/anorex/equiv_classes_hmm.m,
example_models/anorex/fit_model_hier_hmm.m,
example_models/anorex/fit_model_hier_hmm_maineffects.m,
example_models/anorex/fit_model_hier_hmm_time.m,
example_models/anorex/fit_model_hier_hmm_timesq.m,
example_models/anorex/fit_model_hmm.m,
example_models/anorex/link_covariates_to_nodes_hier_hmm_time.m,
example_models/anorex/link_covariates_to_nodes_hier_hmm_timesq.m,
example_models/anorex/loadtime.m,
example_models/anorex/loadtimesamplingdata.m,
example_models/brain/construct_bnet_hmm_theta.m,
example_models/brain/fit_modelbrain_domain_theta.m,
example_models/brain/fit_modelbrain_domain_theta_treat.m,
example_models/brain/fit_modelbrain_hmm.m,
example_models/brain/fit_modelbrain_hmm_domain.m,
example_models/hmm/construct_bnet_hmm.m,
example_models/hmm/fit_model_hmm.m,
example_models/hmm/generate_hmm_data.m,
example_models/hmm/hmm.xls,
example_models/mixed_lltm/construct_bnet_mixlltm.m,
example_models/mixed_lltm/fit_model_mixed_lltm.m

[deep dark]

ss

[ALex Plus]

Wow!

[Haritha Jonnalagadda]

This is very useful.iam doing my project on HMM and i got lot of information in this.I thank a lot.

[Harish Solanki]

This very useful for any one. http://www.freshwarehousing.com/

[ARAVIND]