Merge branch 'master' into master

rust-ml · Aug 7, 2022 · a11b682 · a11b682
2 parents 47aae58 + 870107f
commit a11b682
Show file tree

Hide file tree

Showing 173 changed files with 4,687 additions and 2,044 deletions.
diff --git a/.github/workflows/benching.yml b/.github/workflows/benching.yml
@@ -6,10 +6,6 @@ jobs:
   testing:
     name: benching
     runs-on: ubuntu-18.04
-    if: github.event.pull_request.draft == false
-    container:
-      image: rustmath/mkl-rust:1.43.0
-      options: --security-opt seccomp=unconfined
 
     steps:
       - name: Checkout sources
@@ -26,4 +22,4 @@ jobs:
         uses: actions-rs/cargo@v1
         with:
           command: bench
-          args: iai --all --features intel-mkl-system
+          args: iai --all
diff --git a/.github/workflows/checking.yml b/.github/workflows/checking.yml
@@ -4,20 +4,20 @@ name: Check For Build Errors
 
 jobs:
   check:
-    name: check
-    runs-on: ubuntu-latest
-    if: github.event.pull_request.draft == false
+    name: check-${{ matrix.toolchain }}-${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        toolchain:
-          - 1.54.0
+        toolchain: 
+          - 1.57.0
           - stable
-        experimental: [false]
-        include:
-          - toolchain: nightly
-            experimental: true
-    continue-on-error: ${{ matrix.experimental }}
+          - nightly
+        os:
+          - ubuntu-latest
+          - windows-latest
+
+    continue-on-error: ${{ matrix.toolchain == 'nightly' }}
 
     steps:
       - name: Checkout sources
@@ -30,15 +30,23 @@ jobs:
           toolchain: ${{ matrix.toolchain }}
           override: true
 
+      - name: Log active toolchain
+        run: rustup show
+
+      # Check if linfa compiles by itself without uniting dependency features with other crates
+      - name: Run cargo check on linfa
+        uses: actions-rs/cargo@v1
+        with:
+          command: check
+
       - name: Run cargo check (no features)
         uses: actions-rs/cargo@v1
         with:
           command: check
-          args: --workspace
+          args: --workspace --all-targets
 
       - name: Run cargo check (with serde)
-        if: ${{ matrix.toolchain != '1.54.0' }} # The following args don't work on older versions of Cargo
         uses: actions-rs/cargo@v1
         with:
           command: check
-          args: --workspace --features "linfa-clustering/serde linfa-ica/serde linfa-kernel/serde linfa-reduction/serde linfa-svm/serde linfa-elasticnet/serde linfa-pls/serde linfa-trees/serde linfa-nn/serde"
+          args: --workspace --all-targets --features "linfa-clustering/serde linfa-ica/serde linfa-kernel/serde linfa-reduction/serde linfa-svm/serde linfa-elasticnet/serde linfa-pls/serde linfa-trees/serde linfa-nn/serde"
diff --git a/.github/workflows/codequality.yml b/.github/workflows/codequality.yml
@@ -7,11 +7,10 @@ jobs:
   codequality:
     name: codequality
     runs-on: ubuntu-latest
-    if: github.event.pull_request.draft == false
     strategy:
       matrix:
         toolchain:
-          - 1.54.0
+          - 1.57.0
           - stable
 
     steps:
@@ -42,9 +41,6 @@ jobs:
     name: coverage
     runs-on: ubuntu-18.04
     if: github.event.pull_request.draft == false
-    container:
-      image: rustmath/mkl-rust:1.43.0
-      options: --security-opt seccomp=unconfined
 
     steps:
       - name: Checkout sources
@@ -74,7 +70,7 @@ jobs:
 
       - name: Generate code coverage
         run: |
-          cargo tarpaulin --verbose --features intel-mkl-system --timeout 120 --out Xml --all --release
+          cargo tarpaulin --verbose --timeout 120 --out Xml --all --release
       - name: Upload to codecov.io
         uses: codecov/codecov-action@v1
         with:

diff --git a/.github/workflows/testing.yml b/.github/workflows/testing.yml
@@ -4,18 +4,17 @@ name: Run Tests
 
 jobs:
   testing:
-    name: testing
-    runs-on: ubuntu-18.04
-    if: github.event.pull_request.draft == false
-    container:
-      image: rustmath/mkl-rust:1.43.0
-      options: --security-opt seccomp=unconfined
+    name: testing-${{ matrix.toolchain }}-${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
         toolchain:
-          - 1.54.0
+          - 1.57.0
           - stable
+        os:
+          - ubuntu-18.04
+          - windows-2019
 
     steps:
       - name: Checkout sources
@@ -28,8 +27,38 @@ jobs:
           toolchain: ${{ matrix.toolchain }}
           override: true
 
-      - name: Run cargo test in release mode
+      - name: Run cargo test
         uses: actions-rs/cargo@v1
         with:
           command: test
-          args: --all --release --features intel-mkl-system
+          args: --release --workspace
+
+  testing-blas:
+    name: testing-with-BLAS-${{ matrix.toolchain }}-${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        toolchain:
+          - 1.57.0
+          - stable
+        os:
+          - ubuntu-18.04
+          - windows-2019
+
+    steps:
+      - name: Checkout sources
+        uses: actions/checkout@v2
+
+      - name: Install toolchain
+        uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: ${{ matrix.toolchain }}
+          override: true
+
+      - name: Run cargo test with BLAS enabled
+        uses: actions-rs/cargo@v1
+        with:
+          command: test
+          args: --release --workspace --features intel-mkl-static,linfa-clustering/blas,linfa-ica/blas,linfa-reduction/blas,linfa-linear/blas,linfa-preprocessing/blas,linfa-pls/blas,linfa-elasticnet/blas
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,3 +1,37 @@
+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
 ========================
 

diff --git a/Cargo.toml b/Cargo.toml
@@ -1,10 +1,11 @@
 [package]
 name = "linfa"
-version = "0.5.0"
+version = "0.6.0"
 authors = [
     "Luca Palmieri <rust@lpalmieri.com>",
     "Lorenz Schmidt <bytesnake@mailbox.org>",
-    "Paul Körbitz <koerbitz@google.com>"
+    "Paul Körbitz <koerbitz@google.com>",
+    "Yuhan Lin <yuhanliin@protonmail.com>"
 ]
 description = "A Machine Learning framework for Rust"
 edition = "2018"
@@ -21,14 +22,14 @@ exclude = [".github/"]
 [features]
 default = []
 
-netlib-static = ["blas", "ndarray-linalg", "netlib-src/static"]
-netlib-system = ["blas", "ndarray-linalg", "netlib-src/system"]
+netlib-static = ["blas", "ndarray-linalg/netlib-static"]
+netlib-system = ["blas", "ndarray-linalg/netlib-system"]
 
-openblas-static = ["blas", "ndarray-linalg", "openblas-src/static"]
-openblas-system = ["blas", "ndarray-linalg", "openblas-src/system"]
+openblas-static = ["blas", "ndarray-linalg/openblas-static"]
+openblas-system = ["blas", "ndarray-linalg/openblas-system"]
 
-intel-mkl-static = ["blas", "ndarray-linalg", "intel-mkl-src/mkl-static-lp64-seq", "intel-mkl-src/download"]
-intel-mkl-system = ["blas", "ndarray-linalg", "intel-mkl-src/mkl-dynamic-lp64-seq"]
+intel-mkl-static = ["blas", "ndarray-linalg/intel-mkl-static"]
+intel-mkl-system = ["blas", "ndarray-linalg/intel-mkl-system"]
 
 blas = ["ndarray/blas"]
 
@@ -39,10 +40,12 @@ num-traits = "0.2"
 rand = { version = "0.8", features = ["small_rng"] }
 approx = "0.4"
 
-ndarray = { version = "0.15", default-features = false, features = ["approx"] }
+ndarray = { version = "0.15", features = ["approx"] }
 ndarray-linalg = { version = "0.14", optional = true }
 
 thiserror = "1.0"
+# Temporary fix for compatibility issue between intel-mkl-tool and anyhow (https://github.com/rust-math/intel-mkl-src/issues/68)
+anyhow = { version = "<=1.0.48" }
 
 [dependencies.serde_crate]
 package = "serde"
@@ -51,23 +54,6 @@ version = "1.0"
 default-features = false
 features = ["std", "derive"]
 
-[dependencies.intel-mkl-src]
-version = "0.6.0"
-default-features = false
-optional = true
-
-[dependencies.netlib-src]
-version = "0.8.0"
-optional = true
-features = ["cblas"]
-default-features = false
-
-[dependencies.openblas-src]
-version = "0.10.4"
-optional = true
-default-features = false
-features = ["cblas"]
-
 [dev-dependencies]
 ndarray-rand = "0.14"
 linfa-datasets = { path = "datasets", features = ["winequality", "iris", "diabetes"] }

diff --git a/README.md b/README.md
@@ -36,35 +36,34 @@ Where does `linfa` stand right now? [Are we learning yet?](http://www.arewelearn
 | [elasticnet](algorithms/linfa-elasticnet/) | Elastic Net | Tested | Supervised learning | Linear regression with elastic net constraints |
 | [logistic](algorithms/linfa-logistic/) | Logistic regression | Tested  | Partial fit | Builds two-class logistic regression models
 | [reduction](algorithms/linfa-reduction/) | Dimensionality reduction | Tested | Pre-processing | Diffusion mapping and Principal Component Analysis (PCA) |
-| [trees](algorithms/linfa-trees/) | Decision trees | Experimental  | Supervised learning | Linear decision trees
+| [trees](algorithms/linfa-trees/) | Decision trees | Tested / Benchmarked  | Supervised learning | Linear decision trees
 | [svm](algorithms/linfa-svm/) | Support Vector Machines | Tested  | Supervised learning | Classification or regression analysis of labeled datasets | 
 | [hierarchical](algorithms/linfa-hierarchical/) | Agglomerative hierarchical clustering | Tested | Unsupervised learning | Cluster and build hierarchy of clusters |
 | [bayes](algorithms/linfa-bayes/) | Naive Bayes | Tested | Supervised learning | Contains Gaussian Naive Bayes |
 | [ica](algorithms/linfa-ica/) | Independent component analysis | Tested | Unsupervised learning | Contains FastICA implementation |
 | [pls](algorithms/linfa-pls/) | Partial Least Squares | Tested | Supervised learning | Contains PLS estimators for dimensionality reduction and regression |
 | [tsne](algorithms/linfa-tsne/) | Dimensionality reduction| Tested | Unsupervised learning | Contains exact solution and Barnes-Hut approximation t-SNE |
-| [preprocessing](algorithms/linfa-preprocessing/) |Normalization & Vectorization| Tested | Pre-processing | Contains data normalization/whitening and count vectorization/tf-idf |
+| [preprocessing](algorithms/linfa-preprocessing/) |Normalization & Vectorization| Tested / Benchmarked | Pre-processing | Contains data normalization/whitening and count vectorization/tf-idf |
 | [nn](algorithms/linfa-nn/) | Nearest Neighbours & Distances | Tested / Benchmarked | Pre-processing | Spatial index structures and distance functions |
+| [ftrl](algorithms/linfa-ftrl/) | Follow The Reguralized Leader - proximal | Tested  / Benchmarked | Partial fit | Contains L1 and L2 regularization. Possible incremental update |
 
 We believe that only a significant community effort can nurture, build, and sustain a machine learning ecosystem in Rust - there is no other way forward.
 
 If this strikes a chord with you, please take a look at the [roadmap](https://github.com/rust-ml/linfa/issues/7) and get involved!
 
 ## BLAS/Lapack backend
 
-At the moment you can choose between the following BLAS/LAPACK backends: `openblas`, `netblas` or `intel-mkl`
+Some algorithm crates need to use an external library for linear algebra routines. By default, we use a pure-Rust implementation. However, you can also choose an external BLAS/LAPACK backend library instead, by enabling the `blas` feature and a feature corresponding to your BLAS backend. Currently you can choose between the following BLAS/LAPACK backends: `openblas`, `netblas` or `intel-mkl`.
 
 |Backend  | Linux | Windows | macOS |
 |:--------|:-----:|:-------:|:-----:|
 |OpenBLAS |✔️      |-        |-      |
 |Netlib   |✔️      |-        |-      |
 |Intel MKL|✔️      |✔️        |✔️      |
 
-For example if you want to use the system IntelMKL library for the PCA example, then pass the corresponding feature:
-```
-cd linfa-reduction && cargo run --release --example pca --features linfa/intel-mkl-system
-```
-This selects the `intel-mkl` system library as BLAS/LAPACK backend. On the other hand if you want to compile the library and link it with the generated artifacts, pass `intel-mkl-static`.
+Each BLAS backend has two features available. The feature allows you to choose between linking the BLAS library in your system or statically building the library. For example, the features for the `intel-mkl` backend are `intel-mkl-static` and `intel-mkl-system`.
+
+An example set of Cargo flags for enabling the Intel MKL backend on an algorithm crate is `--features blas,linfa/intel-mkl-system`. Note that the BLAS backend features are defined on the `linfa` crate, and should only be specified for the final executable.
 
 # License
 Dual-licensed to be compatible with the Rust project.

diff --git a/algorithms/linfa-bayes/Cargo.toml b/algorithms/linfa-bayes/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "linfa-bayes"
-version = "0.5.0"
+version = "0.6.0"
 authors = ["VasanthakumarV <vasanth260m12@gmail.com>"]
 description = "Collection of Naive Bayes Algorithms"
 edition = "2018"
@@ -11,12 +11,12 @@ keywords = ["factorization", "machine-learning", "linfa", "unsupervised"]
 categories = ["algorithms", "mathematics", "science"]
 
 [dependencies]
-ndarray = { version = "0.15" , features = ["blas", "approx"]}
+ndarray = { version = "0.15" , features = ["approx"]}
 ndarray-stats = "0.5"
 thiserror = "1.0"
 
-linfa = { version = "0.5.0", path = "../.." }
+linfa = { version = "0.6.0", path = "../.." }
 
 [dev-dependencies]
 approx = "0.4"
-linfa-datasets = { version = "0.5.0", path = "../../datasets", features = ["winequality"] }
+linfa-datasets = { version = "0.6.0", path = "../../datasets", features = ["winequality"] }