Changelog History
Page 1
-
v4.0.0 Changes
????-??-?? Add Multiple Pole Balancing Environment (#1901, #1951).
Added functionality for scaling of data (#1876); see the command-line binding
mlpack_preprocess_scale
or Python bindingpreprocess_scale()
.Add new parameter
maximum_depth
to decision tree and random forest bindings (#1916).Fix prediction output of softmax regression when test set accuracy is calculated (#1922).
Action struct in continuous RL environments now stores the action as a
double
instead ofdouble[1]
(#1941, #1931).Pendulum environment now checks for termination. All RL environments now have an option to terminate after a set number of time steps (no limit by default) (#1941).
Add support for probabilistic KDE (kernel density estimation) error bounds when using the Gaussian kernel (#1934).
-
v3.4.2 Changes
October 28, 2020๐ Released Oct. 28, 2020.
- โ Added Mean Absolute Percentage Error.
- โ Added Softmin activation function as layer in ann/layer.
- Fix spurious ARMA_64BIT_WORD compilation warnings on 32-bit systems (#2665).
-
v3.4.1 Changes
September 07, 2020 -
v3.4.0 Changes
September 01, 2020๐ Released Sept. 1st, 2020.
โ Issue warnings when metrics produce NaNs in KFoldCV (#2595).
Added bindings for R during Google Summer of Code (#2556).
โ Added common striptype function for all bindings (#2556).
๐จ Refactored common utility function of bindings to bindings/util (#2556).
Renamed InformationGain to HoeffdingInformationGain in
methods/hoeffding_trees/information_gain.hpp
(#2556).โ Added macro for changing stream of printing and warnings/errors (#2556).
โ Added Spatial Dropout layer (#2564).
๐ฎ Force CMake to show error when it didn't find Python/modules (#2568).
๐จ Refactor
ProgramInfo()
to separate out all the different information (#2558).โ Add bindings for one-hot encoding (#2325).
โ Added Soft Actor-Critic to RL methods (#2487).
โ Added Categorical DQN to q_networks (#2454).
โ Added N-step DQN to q_networks (#2461).
โ Add Silhoutte Score metric and Pairwise Distances (#2406).
โ Add Go bindings for some missed models (#2460).
Replace boost program_options dependency with CLI11 (#2459).
โ Additional functionality for the ARFF loader (#2486); use case sensitive categories (#2516).
Add
bayesian_linear_regression
binding for the command-line, Python, Julia, and Go. Also called "Bayesian Ridge", this is equivalent to a version of linear regression where the regularization parameter is automatically tuned (#2030).๐ Fix defeatist search for spill tree traversals (#2566, #1269).
๐ Fix incremental training of logistic regression models (#2560).
Change default configuration of
BUILD_PYTHON_BINDINGS
toOFF
(#2575). -
v3.3.2
June 18, 2020 -
v3.3.1 Changes
April 30, 2020๐ Released April 29th, 2020.
๐ Minor Julia and Python documentation fixes (#2373).
โก๏ธ Updated terminal state and fixed bugs for Pendulum environment (#2354, #2369).
โ Added
EliSH
activation function (#2323).โ Add L1 Loss function (#2203).
Pass CMAKE_CXX_FLAGS (compilation options) correctly to Python build (#2367).
๐ฆ Expose ensmallen Callbacks for sparseautoencoder (#2198).
๐ Bugfix for LARS class causing invalid read (#2374).
โ Add serialization support from Julia; use
mlpack.serialize()
andmlpack.deserialize()
to save and load fromIOBuffer
s. -
v3.3.0 Changes
April 07, 2020๐ Released April 7th, 2020.
Templated return type of
Forward function
of loss functions (#2339).โ Added
R2 Score
regression metric (#2323).โ Added
mean squared logarithmic error
loss function for neural networks (#2210).โ Added
mean bias loss function
for neural networks (#2210).๐ The DecisionStump class has been marked deprecated; use the
DecisionTree
class withNoRecursion=true
or useID3DecisionStump
instead (#2099).โ Added
probabilities_file
parameter to get the probabilities matrix of AdaBoost classifier (#2050).๐ Fix STB header search paths (#2104).
โ Add
DISABLE_DOWNLOADS
CMake configuration option (#2104).โ Add padding layer in TransposedConvolutionLayer (#2082).
๐ Fix pkgconfig generation on non-Linux systems (#2101).
๐ฒ Use log-space to represent HMM initial state and transition probabilities (#2081).
โ Add functions to access parameters of
Convolution
andAtrousConvolution
layers (#1985).โ Add Compute Error function in lars regression and changing Train function to return computed error (#2139).
Add Julia bindings (#1949). Build settings can be controlled with the
BUILD_JULIA_BINDINGS=(ON/OFF)
andJULIA_EXECUTABLE=/path/to/julia
CMake parameters.CMake fix for finding STB include directory (#2145).
Add bindings for loading and saving images (#2019);
mlpack_image_converter
from the command-line,mlpack.image_converter()
from Python.โ Add normalization support for CF binding (#2136).
โ Add Mish activation function (#2158).
โก๏ธ Update
init_rules
in AMF to allow users to merge two initialization rules (#2151).โ Add GELU activation function (#2183).
๐ Better error handling of eigendecompositions and Cholesky decompositions (#2088, #1840).
โ Add LiSHT activation function (#2182).
โ Add Valid and Same Padding for Transposed Convolution layer (#2163).
โ Add CELU activation function (#2191)
โ Add Log-Hyperbolic-Cosine Loss function (#2207)
๐ Change neural network types to avoid unnecessary use of rvalue references (#2259).
โฌ๏ธ Bump minimum Boost version to 1.58 (#2305).
๐จ Refactor STB support so
HAS_STB
macro is not needed when compiling against mlpack (#2312).โ Add Hard Shrink Activation Function (#2186).
โ Add Soft Shrink Activation Function (#2174).
โ Add Hinge Embedding Loss Function (#2229).
โ Add Cosine Embedding Loss Function (#2209).
โ Add Margin Ranking Loss Function (#2264).
๐ Bugfix for incorrect parameter vector sizes in logistic regression and softmax regression (#2359).
-
v3.2.2 Changes
November 26, 2019๐ Released Nov. 26th, 2019.
- โ Add
valid
andsame
padding option inConvolution
andAtrous Convolution
layer (#1988). - โ Add Model() to the FFN class to access individual layers (#2043).
- ๐ Update documentation for pip and conda installation packages (#2044).
- Add bindings for linear SVM (#1935);
mlpack_linear_svm
from the command-line,linear_svm()
from Python. - โ Add support to return the layer name as
std::string
(#1987). - Speed and memory improvements for the Transposed Convolution layer (#1493).
- ๐ Fix Windows Python build configuration (#1885).
- Validate md5 of STB library after download (#2087).
- ๐ Add
__version__
to__init__.py
(#2092).
- โ Add
-
v3.2.1
October 01, 2019 -
v3.2.0 Changes
September 26, 2019๐ Released Sept. 25, 2019.
๐ Fix occasionally-failing RADICAL test (#1924).
๐ Fix gcc 9 OpenMP compilation issue (#1970).
โ Added support for loading and saving of images (#1903).
โ Add Multiple Pole Balancing Environment (#1901, #1951).
Added functionality for scaling of data (#1876); see the command-line binding
mlpack_preprocess_scale
or Python bindingpreprocess_scale()
.โ Add new parameter
maximum_depth
to decision tree and random forest bindings (#1916).๐ Fix prediction output of softmax regression when test set accuracy is calculated (#1922).
0๏ธโฃ Pendulum environment now checks for termination. All RL environments now have an option to terminate after a set number of time steps (no limit by default) (#1941).
โ Add support for probabilistic KDE (kernel density estimation) error bounds when using the Gaussian kernel (#1934).
๐ Fix negative distances for cover tree computation (#1979).
๐ Fix cover tree building when all pairwise distances are 0 (#1986).
๐ Improve KDE pruning by reclaiming not used error tolerance (#1954, #1984).
๐ Optimizations for sparse matrix accesses in z-score normalization for CF (#1989).
Add
kmeans_max_iterations
option to GMM training bindinggmm_train_main
.โฌ๏ธ Bump minimum Armadillo version to 8.400.0 due to ensmallen dependency requirement (#2015).