Releases: JuliaGaussianProcesses/KernelFunctions.jl
Releases · JuliaGaussianProcesses/KernelFunctions.jl
v0.4.3
KernelFunctions v0.4.3
Closed issues:
Merged pull requests:
- Fixing some documentation bugs (#119) (@theogf)
- Apply Transform on vector of vectors (#129) (@theogf)
- CompatHelper: bump compat for "Documenter" to "0.25" for package docs (#132) (@github-actions[bot])
- CompatHelper: bump compat for "Flux" to "0.11" for package test (#136) (@github-actions[bot])
- Add datatype for multi-output GP input (#138) (@sharanry)
- Tweak implementation (#140) (@willtebbutt)
v0.4.2
v0.4.1
KernelFunctions v0.4.1
Closed issues:
- Add Tests (#8)
- Support for non-Euclidean domains (#10)
- obsdim (#43)
- Get rid of
_kernel(#72) - kappa vs _kernel vs k(x, y) (#87)
- pairwise! for kernelmatrix! (#90)
- Remove
_kernelmatrix(#91) - Replace reduce by mapreduce for kernelproduct (#92)
- ScaledKernel (#94)
- _scale (#96)
- LowRankTransform (#97)
Merged pull requests:
- Add NeuralNetOneKernel (#70) (@sharanry)
- Add i-times integrated Wiener Kernel (#77) (@sharanry)
- Add Spectral Mixture Kernel (#80) (@sharanry)
- Remove _kernel, sort out kappa (#95) (@willtebbutt)
- Move to AbstractVector (#99) (@willtebbutt)
- Use pairwise! for inplace operations (#102) (@devmotion)
- Simplify implementation of
TensorProduct(#103) (@devmotion) - Optimization of ScaleTransform for SimpleKernel (#104) (@devmotion)
- Use LinearTransform instead of LowRankTransform (#106) (@devmotion)
- Rewrite Zygote tests without broadcasting (#110) (@devmotion)
- Length check before applying TensorProduct (#111) (@sharanry)
- Series of tests for AD (#114) (@theogf)
- Remove SqEuclidean roundoff hack (#117) (@devmotion)
- CompatHelper: bump compat for "PDMats" to "0.10" for package test (#120) (@github-actions[bot])
- Some fixes for MaternKernel (#121) (@devmotion)
- Patch positive value for test on ScaledKernel (#122) (@theogf)
- CompatHelper: bump compat for "Zygote" to "0.5" for package test (#124) (@github-actions[bot])
- Relaxed transform to be used on Kernel (#125) (@theogf)
v0.4.0
KernelFunctions v0.4.0
Closed issues:
- Add periodic kernels (#34)
- Objects of type KernelProduct are not callable (#55)
- Split packages between methods and collection of kernels (#86)
Merged pull requests:
- Add Periodic Kernel (#50) (@theogf)
- Add Gabor Kernel (#52) (@sharanry)
- Add Piecewise Polynomial Kernel (#76) (@sharanry)
- Add general TensorProduct kernel (#81) (@devmotion)
- Separate directory for base kernels (#82) (@devmotion)
- [WIP] Rework on kernelmatrix to work with Vectors and more complex kernels (#83) (@theogf)
- Added ColVecs (#84) (@theogf)
- Improved docs, printing and testing (#85) (@theogf)
- Added VecOfVecs and RowVecs (#89) (@theogf)
v0.3.2
KernelFunctions v0.3.2
Closed issues:
- Moving KernelFunctions.jl to the GPMLj organisation (#46)
Merged pull requests:
- Add Cosine Kernel (#45) (@sharanry)
- Add Fractional Brownian motion kernel (#48) (@sharanry)
- Add non-unicode keywords aliases (#49) (@theogf)
- Add alias to WhiteKernel, EyeKernel (#51) (@sharanry)
- Stop testing on 1.2, start on 1.4 (#53) (@willtebbutt)
- Add Mahalanobis distance-based kernel (#54) (@sharanry)
- Created concrete types to call syntactic sugar on all kernels (#57) (@theogf)
- CompatHelper: bump compat for "StatsBase" to "0.33" (#58) (@github-actions[bot])
- Update links (#59) (@devmotion)
- Fix another link (#60) (@devmotion)
- Do not export params (#61) (@devmotion)
- Remove redundant definition (#62) (@devmotion)
- Refactors the test suite (#63) (@willtebbutt)
- Corrected docs and improved formatting (#66) (@theogf)
- Update Travis configuration (#69) (@devmotion)
v0.3.1
KernelFunctions v0.3.1
Closed issues:
- Create a trainable function for Flux compatibility (#31)
Merged pull requests:
- Fix PDMats warning (#40) (@devmotion)
- params(), Flux/Zygote style (#41) (@theogf)
- Fix typo (#42) (@willtebbutt)
v0.3.0
KernelFunctions v0.3.0
- Breaking changes
- calling a transform or a scale in a Kernel constructor is not valid anymore (Better constructors will come soon):
SqExponential(3.0) # Not valid anymore
transform(SqExponential(),3.0) # Right syntax- Kernels with parameters (
ConstantKernelfor example) needs their parameters to be given via a keyword, see the documentation for each kernel
ConstantKernel(c=2.0)- Nystrom approximation : The nystrom approximation has now been adapted from MLKernels.jl
- PDMats and Kronecker : it is now possible to create PDMat matrices and Kronecker matrices, by first using PDMat or Kronecker in a script. See the docs as well
- The transform field has been removed to be contained in
TransformedKernel
Closed issues:
- Type coherence (#3)
- Transformations of kernels (#17)
- Output Types (#18)
- Nystrom approximations (#19)
- Remove adjoints for SqEuclidean and Euclidean (#37)
Merged pull requests:
- CompatHelper: bump compat for "StatsFuns" to "0.9" (#23) (@github-actions[bot])
- CompatHelper: bump compat for "SpecialFunctions" to "0.10" (#24) (@github-actions[bot])
- CompatHelper: bump compat for "Compat" to "3.2" (#25) (@github-actions[bot])
- Remove T parameter from Kernel (#26) (@willtebbutt)
- Fix StackOverflowError (#27) (@devmotion)
- Fix Travis link (#28) (@devmotion)
- Ease implementation of custom kernels (#29) (@devmotion)
- Nystrom Approximation (#30) (@IsakFalk)
- Removing transform field and creating TransformedKernel (and ScaledKernel) (#32) (@theogf)
- CompatHelper: add new compat entry for "StatsBase" at version "0.32" (#33) (@github-actions[bot])
- Install TagBot as a GitHub Action (#35) (@JuliaTagBot)
- Implemented Kronecker matrices with Kronecker.jl (#36) (@theogf)
- Remove adjoints of Euclidean and SqEuclidean (#39) (@devmotion)
v0.2.4
v0.2.4
This tag was signed with the committer’s verified signature.
JuliaTagBot
Julia TagBot
v0.2.3
v0.2.3
This tag was signed with the committer’s verified signature.
JuliaTagBot
Julia TagBot
v0.2.2
v0.2.2
This tag was signed with the committer’s verified signature.
JuliaTagBot
Julia TagBot