v1.0.6

HMMBase v1.0.6

Diff since v1.0.5

Merged pull requests:

• CompatHelper: bump compat for "Clustering" to "0.14" (#19) (@github-actions[bot])

v1.0.5

HMMBase v1.0.5

Diff since v1.0.4

Merged pull requests:

• CompatHelper: bump compat for "ArgCheck" to "2.0" (#15) (@github-actions[bot])
• CompatHelper: bump compat for "Hungarian" to "0.6" (#17) (@github-actions[bot])
• CompatHelper: bump compat for "Distributions" to "0.23" (#18) (@github-actions[bot])

v1.0.4

HMMBase v1.0.4

Utilities

• Add remapseq

Diff since v1.0.3

v1.0.3

HMMBase v1.0.3

Diff since v1.0.2

Merged pull requests:

• CompatHelper: bump compat for "Distributions" to "0.22" (#14) (@github-actions[bot])

v1.0.2

Fixed

• #12 - Error when sampling multivariate HMMs

Development

• Added CompatHelper and TagBot actions

v1.0.1

• Loosen compatibility requirements

HMMBase v1.0

Improvements:

• Faster and cleaner implementations of the forward, backward, Viterbi and EM algorithms.
• New methods: randtransmat, istransmat, statsdists, permute, copy, size, ==
• Internal refactoring, reduced memory allocations

Breaking changes:

• HMM struct renaming: π0, π, D become a, A, B
• Removal of StaticHMM and StaticArrays dependency
• Methods renaming, see documentation for a full list
• Forward/Backward algorithms uses likelihood by default (instead of log-likelihoods), use the logl option to use log-likelihoods
• Baum-Welch algorithm returns hmm, history instead of hmm, logtot
• rand(hmm, T) returns y instead of z, y by default, use seq = true to get z, y

See Migration to v1.0 for details on migrating your code to the new version.

Thanks to @nantonel PR#6 for the many improvements.

v0.0.11

• Performance improvements.
• Initial implementation of the Baum-Welch algorithm.

v0.0.10

https://discourse.julialang.org/t/ann-hmmbase-jl-a-lightweight-and-efficient-hidden-markov-model-abstraction/21604