v0.3.0

ParametrisedConvexApproximators v0.3.0

Diff since v0.2.2

Merged pull requests:

• v0.3.0 (#63) (@JinraeKim)
• Fixing compat (#64) (@JinraeKim)

Closed issues:

• Make the approximators autodiff-compatible (#13)
• Add EPLSE (#38)
• Incompatibility with Flux (#59)

v0.2.2

ParametrisedConvexApproximators v0.2.2

Diff since v0.2.1

Merged pull requests:

• Hotfix for NormalisedApproximator (#62) (@JinraeKim)

v0.2.1

ParametrisedConvexApproximators v0.2.1

Diff since v0.2.0

Merged pull requests:

• Hotfix/compatibility flux (#60) (@JinraeKim)
• v0.2.1 (#61) (@JinraeKim)

v0.2.0

ParametrisedConvexApproximators v0.2.0

Diff since v0.1.1

Closed issues:

• Change the package and function names (British to American) (#20)
• Add basic test files (#21)
• Add CI (continuous integration) (#22)
• Add a type for trainer (#24)
• Add data split for "train, validate, test" (#25)
• Add dataset (#29)
• Save model after training (best/last) (#33)
• Add a modified PLSE with zero-gradient function (#36)
• Add more simple functions including "non parameterized convex" ones (#39)
• Add visualization (loss graph) (#40)
• Add data normalization (#41)
• Add DLSE (#42)
• Change how to construct MA and LSE (#47)
• Bug fix of DLSE; not trained (#48)
• Normalized approximator for trainer (#52)
• Dataset save and load (#55)
• Should we replace the sample and training part with Surrogates.jl? (#56)

Merged pull requests:

• Add arXiv (#15) (@JinraeKim)
• Add ci (#23) (@JinraeKim)
• v0.2.0 (#26) (@JinraeKim)
• Add split_data2 and split_data3 (#27) (@JinraeKim)
• Deprecate UnPack (#28) (@JinraeKim)
• Add SimpleDataset (#30) (@JinraeKim)
• Adding supervised learning trainer (#31) (@JinraeKim)
• British to American (#32) (@JinraeKim)
• Update dataset; add an example of save and load network model (#34) (@JinraeKim)
• Add a modified PLSE; strict (#37) (@JinraeKim)
• Add loss graph (#43) (@JinraeKim)
• Add more simple functions (#44) (@JinraeKim)
• Add DLSE (#46) (@JinraeKim)
• Bux fix dlse (#49) (@JinraeKim)
• Change the way how to construct the MA and LSE (#50) (@JinraeKim)
• Add NormalizedApproximator and MaxAbsNormalizedApproximator (#51) (@JinraeKim)
• Better initial guess (#53) (@JinraeKim)
• Normalized network for trainer (#54) (@JinraeKim)
• Update README.md (#57) (@JinraeKim)
• Update Project.toml for compat (#58) (@JinraeKim)

v0.1.1

ParametrisedConvexApproximators v0.1.1

Merged pull requests:

• Add basic form of approximators and visualisation tools (#4) (@JinraeKim)
• Add to-do list (#5) (@JinraeKim)
• Add normalisation (#7) (@JinraeKim)
• Rewrite code (#12) (@JinraeKim)