oneAPI Data Analytics Library (oneDAL)
The release IntelĀ® oneAPI Data Analytics Library 2023.1 introduces the following changes:
The release IntelĀ® oneAPI Data Analytics Library 2021.7.1 introduces the following changes:
zlib and bzip2 methods of compression were deprecated. They are dispatched to the lzo method starting this versioneigenvectors, eigenvalues, variances and means) and precomputed method for PCA algorithm.The release IntelĀ® oneAPI Data Analytics Library 2021.6 introduces the following changes:
Kaggle kernels for IntelĀ® Extension for Scikit-learn:
sendrecv_replace communicator methodThe release introduces the following changes:
The following additional materials were created:
oneDAL samples:
IntelĀ® Extension for Scikit-learn samples:
daal4py samples:
Kaggle kernels for IntelĀ® Extension for Scikit-learn:
The release introduces the following changes:
The following additional materials were created:
Medium blogs:
Anaconda blogs:
Oracle blogs:
Kaggle kernels:
Added demo samples comparing the usage of IntelĀ® Extension for Scikit-learn and the original Scikit-learn for KNN, Logistic Regression, SVM and Random Forest algorithms
python -m sklearnex.glob patch_sklearn
from sklearnex import patch_sklearn
patch_sklearn(global_patch=True)
dpctl.tensor.usm_ndarray for input and output)set_config and get_config methods. Added the support of target_offload and allow_fallback_to_host options for device offloading scenariospredict_proba in RandomForestClassifier estimatorSVR algorithm trainingNuSVC and NuSVR algorithms trainingRandomForestRegression and RandomForestClassifier algorithms training and predictionKMeans
Random Forest algorithm caused by the exclusion of constant features.NuSVC Multiclass.KMeans convergence inconsistency.train_test_split with specific subset sizes.SVM.The release introduces the following changes:
The following additional materials were created:
Medium blogs:
Kaggle kernels:
Samples that illustrate the usage of Intel Extension for Scikit-learn
pip install scikit-learn-intelex
conda install scikit-learn-intelex -c conda-forge
from sklearnex import patch_sklearn
patch_sklearn()
patch_sklearn for all algorithmspatch_sklearn to patch both fit and predict methods of Logistic Regression when the algorithm is given as a single parameter to patch_sklearn
import sysĀ
import osĀ
import siteĀ
sys.path.append(os.path.join(os.path.dirname(site.getsitepackages()[0]), "site-packages"))Ā
The release introduces the following changes:
pip install daal4py
The following additional materials were created:
roc_auc_score functionLinearRegression, Ridge, SVC, KMeans, PCA, Lasso, ElasticNet, tSNE, KNeighborsClassifier, KNeighborsRegressor, NearestNeighbors, RandomForestClassifier, RandomForestRegressor
RandomForestClassifier and RandomForestRegressor scikit-learn estimators: training and predictionprobability==True parameter: training and predictionScikit-learn patching:
RandomForestClassifier and RandomForestRegressor scikit-learn estimatorspairwise_distances
patch_sklearn and unpatch_sklearn functionsfloat32 or float64 data types. Scikit-learn patching now works with all numpy data types.DataFrame from pandas was used as an input typedaal4py:
oneDAL:
vars.sh script does not support kornShellThe release contains all functionality of IntelĀ® DAAL. See IntelĀ® DAAL release notes for more details.
IntelĀ® Data Analytics Acceleration Library to IntelĀ® oneAPI Data Analytics Library and changed the package names to reflect this.OpenCL and Level Zero backends.Unified Shared Memory (USM) supportK-means, Covariance, PCA, Logistic Regression, Linear Regression, Random Forest Classification and Regression, Gradient Boosting Classification and Regression, kNN, SVM, DBSCAN and Low-order moments
Covariance, PCA, Linear Regression and Low-order moments
Data Management functionality to support DPC++ APIs: a new table type for representation of SYCL-based numeric tables (SyclNumericTable) and an optimized CSV data source
Logistic Regression training and predictionk-Nearest Neighbors prediction with Brute Force methodLogistic Loss and Cross Entropy objective functions
undirected_adjacency_array_graph), where vertex indices can only be of type int32
Jaccard Similarity Coefficients for all pairs of vertices, a batch algorithm that processes the graph by blocksK-means, PCA, kNN
NearestNeighbors and KNeighborsRegressor scikit-learn estimators with Brute Force and K-D tree methodsTSNE scikit-learn estimatorDBSCAN, K-means, Linear and Logistic Regression
LogisticRegression fit, predict and predict_proba methodsKNeighborsClassifier predict, predict_proba and kneighbors methods with ābruteā methodIntelĀ® oneDAL DPC++ APIs does not work on GEN12 graphics with OpenCL backend. Use Level Zero backend for such cases.train_test_split in daal4py patches for Scikit-learn can produce incorrect shuffling on Windows*k-Nearest Neighbors classification algorithm, which for datasets with more than 13 features demonstrates a better performance than the existing K-D tree methodk-Nearest Neighbors search for K-D tree and Brute Force methods with computation of distances to nearest neighbors and their indicesk-Nearest Neighbors classification and search: based on inverse-distance and uniform weightingDecision Forest classification and regression: minObservationsInSplitNode, minWeightFractionInLeafNode, minImpurityDecreaseInSplitNode, maxLeafNodes with best-first strategy and sample weightsSVM) decision function for Multi-class ClassifierSVM training and predictionDecision Forest classification trainingRBF and Linear kernel functionsXGBoost* and LightGBM* models into a daal4py Gradient Boosted Trees model for fast predictionModin* DataFrame as an inputKNeighborsClassifier scikit-learn estimator with Brute Force and K-D tree methodsRandomForestClassifier and RandomForestRegressor scikit-learn estimatorsKMeans and Support Vector Classification (SVC) scikit-learn estimatorsSVC scikit-learn estimatorLasso and ElasticNet scikit-learn estimatorstrain_test_split()
SVC) fit and prediction