probml/dynamax issues and pull requests

#291 - add support for storing just means or covs during kalman filtering

Issue - State: open - Opened by murphyk about 2 years ago

#290 - NaN values in posterior probabilities with sparse transition matrices

Issue - State: closed - Opened by emdupre about 2 years ago - 5 comments

#289 - Implementation: specification of filtered means, covariances, and log-likelihood for EKF and UKF

Pull Request - State: closed - Opened by gerdm about 2 years ago - 8 comments

#288 - Optionally extracting intermediate states for `nonlinear_gaussian_ssm` methods

Issue - State: closed - Opened by gerdm about 2 years ago - 1 comment
Labels: question

#287 - support callable parameters for lgssm

Pull Request - State: closed - Opened by xinglong-li about 2 years ago

#286 - Allow time-dependent parameters of LGSSM to be callable

Issue - State: closed - Opened by xinglong-li about 2 years ago

#285 - Add Dirichlet classifier demo and add utility functions.

Pull Request - State: closed - Opened by gileshd about 2 years ago - 6 comments

#284 - Fix minor typo

Pull Request - State: closed - Opened by canyon289 about 2 years ago - 1 comment

#283 - Add spectral methods for HMM learning

Issue - State: open - Opened by slinderman about 2 years ago - 2 comments
Labels: enhancement, help wanted

#282 - Rename `linear_solve` to `psd_solve`.

Pull Request - State: closed - Opened by gileshd about 2 years ago

#281 - Add wrapper to linear solve.

Pull Request - State: closed - Opened by gileshd about 2 years ago - 1 comment

#280 - make nonlinear SSM for text generation

Issue - State: closed - Opened by murphyk about 2 years ago

#279 - implement square root extended Kalman filter

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#278 - Add support for general likelihood

Pull Request - State: closed - Opened by petergchang about 2 years ago

#277 - add support for normalizing flows + kalman filters

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#276 - add block gibbs sampling for HMM learning

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#275 - add support for parameter learning in NLGSSM models

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#274 - add support for normalizing flow emission model for HMM

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#273 - add support for HMM inference with sparse transition matrices

Issue - State: open - Opened by murphyk about 2 years ago - 6 comments
Labels: help wanted

#272 - implement unscented particle filter using dynamax and blackjax

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#271 - implement rao-blackwellised particle filtering using dynamax and blackjax

Issue - State: open - Opened by murphyk about 2 years ago - 9 comments
Labels: help wanted

#270 - Add a helper function for finding permutations that align two state sequences

Pull Request - State: closed - Opened by slinderman about 2 years ago

#269 - Updating documentation

Pull Request - State: closed - Opened by slinderman about 2 years ago - 1 comment

#267 - make LGSSM blocked gibbs work

Issue - State: closed - Opened by murphyk about 2 years ago - 2 comments

#266 - start adding jax typing hooks

Pull Request - State: closed - Opened by murphyk about 2 years ago

#265 - Packaging for PyPI

Pull Request - State: closed - Opened by slinderman about 2 years ago - 1 comment

#264 - allow nonlinear model callables to be time-indexed

Issue - State: open - Opened by murphyk about 2 years ago
Labels: help wanted

#260 - lgssm.fit_em gives radiacally different behavior on CPU vs TPU

Issue - State: open - Opened by murphyk about 2 years ago - 2 comments

#259 - test jax-typing: does it detect incorrectly shaped named tuple assignments?

Issue - State: closed - Opened by murphyk about 2 years ago

#258 - add more type annotations to class args

Issue - State: closed - Opened by murphyk about 2 years ago - 1 comment

#256 - modify SGD training to avoid using generators

Issue - State: open - Opened by murphyk about 2 years ago - 1 comment
Labels: help wanted

#252 - Type signatures added to *GSSM code.

Pull Request - State: closed - Opened by murphyk over 2 years ago - 1 comment

#250 - Subsume GGF and iterated CMGF

Pull Request - State: closed - Opened by petergchang over 2 years ago - 1 comment

#249 - Remove redundant GGF and iterated CMGF

Issue - State: closed - Opened by petergchang over 2 years ago

#241 - Standardize EM and SGD outputs

Issue - State: closed - Opened by slinderman over 2 years ago - 1 comment

#238 - Implement functionality for arbitrary likelihood distributions and add Poisson.

Pull Request - State: closed - Opened by petergchang over 2 years ago - 3 comments

#237 - Implement functionality for arbitrary likelihood distributions for marginal loglikelihood computation.

Issue - State: closed - Opened by petergchang over 2 years ago - 1 comment

#232 - Compare CMGF to memory-buffered SGD

Issue - State: closed - Opened by petergchang over 2 years ago

#229 - expand KF filtering test to compare to explicit MVN

Issue - State: open - Opened by murphyk over 2 years ago - 1 comment
Labels: good first issue

#219 - Add posterior_sample method to CMGF

Issue - State: open - Opened by xinglong-li over 2 years ago
Labels: help wanted

#215 - Implement Decoupled EKF for MLP Training

Issue - State: closed - Opened by petergchang over 2 years ago

#211 - Implement "fading memory" trick to CMGF

Issue - State: closed - Opened by petergchang over 2 years ago

#210 - Add a `LinearGaussianPoissonSSM` class

Issue - State: closed - Opened by petergchang over 2 years ago

#208 - Add accuracy, speed comparison with SGD for CMGF multiclass logistic regression

Issue - State: closed - Opened by petergchang over 2 years ago

#199 - Compare CMGF logreg to PASS-GLM

Issue - State: closed - Opened by petergchang over 2 years ago

#180 - make kmeans scaling plot

Issue - State: closed - Opened by murphyk over 2 years ago

#157 - [ENH] Low-rank Gaussian HMM

Pull Request - State: closed - Opened by emdupre over 2 years ago - 4 comments

#155 - Include GGF as special case of `cond_moments_gaussian_filter`

Issue - State: closed - Opened by petergchang over 2 years ago - 1 comment

#149 - refactor LGSMM class into subclasses

Issue - State: closed - Opened by murphyk over 2 years ago - 5 comments

#145 - HMM M step: do not update parameters if trainable is False

Issue - State: closed - Opened by murphyk over 2 years ago - 1 comment

#143 - refactor fit_sgd to use TFDS instead of passing in entire data and creating our own minibatches

Issue - State: closed - Opened by murphyk over 2 years ago

#142 - refactor fit_stochastic_em to use TFDS instead of pytorch data loader

Issue - State: closed - Opened by murphyk over 2 years ago

#141 - Remove lax.scan in run_sgd

Pull Request - State: closed - Opened by karalleyna over 2 years ago - 1 comment

#137 - replicate GMMHMM classifier example for isolated word recognition

Issue - State: closed - Opened by murphyk over 2 years ago

#136 - Sequential Monte Carlo

Pull Request - State: closed - Opened by slinderman over 2 years ago - 13 comments

#114 - add PF with laplace proposal for poisson regression

Issue - State: closed - Opened by murphyk over 2 years ago

#112 - implement basic particle filtering examples (possibly using blackjax)

Issue - State: open - Opened by murphyk over 2 years ago - 1 comment
Labels: help wanted

#105 - Implement closed E and M steps for Categorical, Multinomial, and PoissonHMM

Issue - State: closed - Opened by slinderman over 2 years ago - 1 comment

#103 - Test PoissonHMM

Issue - State: closed - Opened by karalleyna over 2 years ago

#99 - Make HMM learning work with variable length time series

Issue - State: open - Opened by slinderman over 2 years ago - 6 comments
Labels: help wanted

#98 - Test GMMHMM

Issue - State: closed - Opened by karalleyna over 2 years ago

#97 - Create GMMHMM class

Issue - State: closed - Opened by karalleyna over 2 years ago

#96 - Test GaussianHMM

Issue - State: closed - Opened by karalleyna over 2 years ago

#95 - Add missing tests for Categorical HMM from hmmlearn

Issue - State: closed - Opened by karalleyna over 2 years ago

#94 - Recreate test_base.py

Issue - State: closed - Opened by karalleyna over 2 years ago

#91 - Recreate GaussianHMM demo from hmmlearn

Issue - State: closed - Opened by karalleyna over 2 years ago

#90 - Add features that hmmlearn has

Issue - State: closed - Opened by karalleyna over 2 years ago

#80 - implement ukf/demos/ukf_spiral_2d_demo.py

Issue - State: closed - Opened by murphyk over 2 years ago

#59 - reproduce linderman-lab LDS demos

Issue - State: closed - Opened by murphyk over 2 years ago

#57 - add support for MAP parameter estimation for LDS

Issue - State: closed - Opened by murphyk over 2 years ago - 1 comment

#50 - allow EM to work with variable length sequnces

Issue - State: closed - Opened by murphyk over 2 years ago - 3 comments

#41 - blocked gibbs sampling for LGSSM as alternative to EM

Issue - State: closed - Opened by murphyk over 2 years ago - 2 comments

#21 - implement hidden semi-markov models

Issue - State: open - Opened by murphyk over 2 years ago
Labels: help wanted

#20 - implement non-parametric HMMs

Issue - State: open - Opened by murphyk over 2 years ago
Labels: help wanted

#17 - implement GaussianARHMM

Issue - State: closed - Opened by murphyk over 2 years ago - 1 comment

#13 - implement Laplace-EM algorithm for LGSSM with Poisson likelihoods

Issue - State: open - Opened by murphyk over 2 years ago
Labels: help wanted

#10 - implement lgssm_fixed_lag_smoother

Issue - State: closed - Opened by murphyk over 2 years ago

GitHub / probml/dynamax issues and pull requests