Skip to content

Latest commit

 

History

History
225 lines (179 loc) · 9.21 KB

CHANGELOG.md

File metadata and controls

225 lines (179 loc) · 9.21 KB

Version 0.8.0 - 2025-09-30

New Algorithms

  • Bernouilli naive bayes in [linfa-bayes] by [@wildart]
  • Bootstrap aggregation (aka Bagging) in new ensemble methods crate [linfa-ensemble] by [@jk1015]

Changes

  • add max_features and tokenizer_function to CountVectorizer in linfa-preprocessing
  • add predict_proba() to Gaussian mixture model in linfa-clustering
  • add predict_proba() and predict_log_proba() to algorithms in linfa-bayes
  • add target names to dataset
  • fix SVR parameterization in linfa-svm
  • fix serde support for algorithms in linfa-pls
  • fix confusion matrix: use predicted and ground thruth labels, make it reproducible
  • fix dataset names after shuffling
  • bump ndarray to 0.16, argmin to 0.11.0, kdtree to 0.7.0, statrs to 0.18, sprs to 0.11
  • bump MSRV to 1.87.0

Version 0.7.1 - 2025-01-14

New Algorithms

  • Random Projection added to linfa-reduction by [@GBathie]

Changes

  • add serde support to linfa-clustering
  • add accessors for classes in linfa-logistics
  • add accessors for Pca attributes in linfa-reduction
  • add wasm-bindgenfeature to use linfa in the browser
  • fix covariance update for GaussianMixtureModel in linfa-clustering
  • bump ndarray-linalg to 0.16 and argmin to 0.9.0
  • bump MSRV to 1.71.1

Version 0.7.0 - 2023-10-15

Changes

  • add array_from_gz_csv and array_from_csv in linfa-datasets
  • make Serde support in linfa-linear, linfa-logistic, and linfa-ftrl optional
  • bump argmin to 0.8.1
  • add Serde support to linfa-preprocessing and linfa-bayes
  • make licenses follow SPDX 2.1 license expression standard

Removals

  • Removed Approximate DBSCAN from linfa-clustering due to performance issues. It's now an alias to regular DBSCAN.
  • Removed partitions dependency, which breaks in current versions of Rust.

Version 0.6.1 - 2022-12-03

New Algorithms

  • Isotonic regression added to linfa-linear by [@wildart]
  • Mean absolute percentage error (MAPE) added to linfa by [@bernado-sb]

Changes

  • add constructor for LpDist
  • add Send + Sync to trait objects returned by linfa-nn
  • remove anyhow <= 1.0.48 version restriction
  • bump ndarray to 0.15
  • fix serde support for LogisticRegression

Version 0.6.0 - 2022-06-15

New Algorithms

  • Multinomial Naive Bayes added to linfa-bayes by [@sgrigory]
  • Follow The Regularized Leader added as linfa-ftrl by [@MilaKyr]

Changes

  • BLAS backend no longer required to build Linfa
  • remove SeedableRng trait bound from KMeans and GaussianMixture
  • replace uses of Isaac RNG with Xoshiro RNG

Breaking Changes

  • parametrize AsTargets by the dimensionality of the targets and introduce AsSingleTargets and AsMultiTargets
  • Dataset and DatasetView can now be parametrized by target dimensionality, with 2D being the default
  • single-target algorithms no longer accept 2D target arrays as input
  • cross_validate changed to cross_validate_single, cross_validate_multi changed to cross_validate
  • Pr has been constrained to 0. <= prob <= 1. with new constructors

Version 0.5.1 - 2022-02-28

Note that the commits for this release are in the 0-5-1 branch.

Changes

  • remove Float trait bound from many Dataset impls, making non-float datasets usable
  • fix build errors in 0.5.0 caused by breaking minor releases from dependencies
  • fix bug in k-means where the termination condition of the algorithm was calculated incorrectly
  • fix build failure when building linfa alone, caused by incorrect feature selection for ndarray

Version 0.5.0 - 2021-10-20

New Algorithms

  • Nearest neighbour algorithms and traits have been added as linfa-nn by [@YuhanLiin]
  • OPTICS has been added to linfa-clustering by @xd009642
  • Multinomial logistic regression has been added to linfa-logistic by [@YuhanLiin]

Changes

  • use least squares solver from ndarray-linalg in linfa-linear (3dc9cb0)
  • optimized DBSCAN by replacing linear range query implementation with KD-tree (44f91d0)
  • allow distance metrics other than Euclidean to be used for KMeans (4e58d8d)
  • enable models to write prediction results into existing memory without allocating (37bc25b)
  • bumped ndarray version to 0.15 and reduced duplicated dependencies (603f821)
  • introduce ParamGuard trait to algorithm parameter sets to enable both explicit and implicit parameter checking (01f912a)
  • replace uses of HNSW with linfa-nn (208a762)

Version 0.4.0 - 2021-04-28

New Algorithms

  • Partial Least Squares Regression has been added as linfa-pls by [@relf]
  • Barnes-Hut t-SNE wrapper has been added as linfa-tsne by [@frjnn]
  • Count-vectorizer and IT-IDF normalization has been added as linfa-preprocessing by [@Sauro98]
  • Platt scaling has been added to linfa-svm by [@bytesnake]
  • Incremental KMeans and KMeans++ and KMeans|| initialization methods added to linfa-clustering by [@YuhanLiin]

Changes

  • bumped ndarray version to 0.14 (8276bdc)
  • change trait signature of linfa::Fit to return Result (a5a479f)
  • add cross_validate to perform K-folding (a5a479f)

Version 0.3.1 - 2021-03-11

In this release of Linfa the documentation is extended, new examples are added and the functionality of datasets improved. No new algorithms were added.

The meta-issue #82 gives a good overview of the necessary documentation improvements and testing/documentation/examples were considerably extended in this release.

Further new functionality was added to datasets and multi-target datasets are introduced. Bootstrapping is now possible for features and samples and you can cross-validate your model with k-folding. We polished various bits in the kernel machines and simplified the interface there.

The trait structure of regression metrics are simplified and the silhouette score introduced for easier testing of K-Means and other algorithms.

Changes

  • improve documentation in all algorithms, various commits
  • add a website to the infrastructure (c8acc785b)
  • add k-folding with and without copying (b0af80546f8)
  • add feature naming and pearson's cross correlation (71989627f)
  • improve ergonomics when handling kernels (1a7982b973)
  • improve TikZ generator in linfa-trees (9d71f603bbe)
  • introduce multi-target datasets (b231118629)
  • simplify regression metrics and add cluster metrics (d0363a1fa8ef)

Version 0.3.0 - 2021-01-21

New Algorithms

  • Approximated DBSCAN has been added to linfa-clustering by [@Sauro98]
  • Gaussian Naive Bayes has been added to linfa-bayes by [@VasanthakumarV]
  • Elastic Net linear regression has been added to linfa-elasticnet by [@paulkoerbitz] and [@bytesnake]

Changes

  • Added benchmark to gaussian mixture models (a3eede55)
  • Fixed bugs in linear decision trees, added generator for TiKZ trees (bfa5aebe7)
  • Implemented serde for all crates behind feature flag (4f0b63bb)
  • Implemented new backend features (7296c9ec4)
  • Introduced linfa-datasets for easier testing (3cec12b4f)
  • Rename Dataset to DatasetBase and introduce Dataset and DatasetView (21dd579cf)
  • Improve kernel tests and documentation (8e81a6d)

Version 0.2.0 - 2020-11-26

New algorithms

  • Ordinary Linear Regression has been added to linfa-linear by [@Nimpruda] and [@paulkoerbitz]
  • Generalized Linear Models has been added to linfa-linear by [VasanthakumarV]
  • Linear decision trees were added to linfa-trees by [@mossbanay]
  • Fast independent component analysis (ICA) has been added to linfa-ica by [@VasanthakumarV]
  • Principal Component Analysis and Diffusion Maps have been added to linfa-reduction by [@bytesnake]
  • Support Vector Machines has been added to linfa-svm by [@bytesnake]
  • Logistic regression has been added to linfa-logistic by [@paulkoerbitz]
  • Hierarchical agglomerative clustering has been added to linfa-hierarchical by [@bytesnake]
  • Gaussian Mixture Models has been added to linfa-clustering by [@relf]

Changes

  • Common metrics for classification and regression have been added
  • A new dataset interface simplifies the work with targets and labels
  • New traits for Transformer, Fit and IncrementalFit standardizes the interface
  • Switched to Github Actions for better integration

Version 0.1.3

New algorithms

  • The DBSCAN clustering algorithm has been added to linfa-clustering (#12 by @xd009642)

Version 0.1.2 (2019-11-25)

New algorithms

  • First release of linfa-clustering:v0.1.0 with the KMeans algorithm (by @LukeMathWalker)
  • First (real) release of linfa, re-exporting linfa-clustering (by @LukeMathWalker)