Intel(R) DAAL 2019 Update 5 is now available. Intel DAAL 2019 packages are now ready for download. Intel DAAL is available as part of the Intel(R) Parallel Studio XE and Intel(R) System Studio. Please visit the https://software.intel.com/en-us/intel-daal.
What's New in Intel(R) DAAL 2019 Update 5:
the New algorithms were added:
DBSCAN (Density-Based Spatial Clustering of Applications with Noise), available in batch and distributed computation modes. Brute-force method for neighborhood computation is implemented.
LASSO (Least Absolute Shrinkage and Selection Operator) regression algorithm with L1 regularization in batch computation mode. An algorithm was designed to support various optimization solvers that handle non-smooth functions.
Coordinate Descent optimization solver in batch computation mode which handles non-smooth functions.
Model convertors for SVM, Gradient Boosted Trees, Decision Forest Classification, Multiclass Classification, Linear Regression and Logistic Regression algorithms. This new functionality allows using Intel® DAAL prediction without Intel® DAAL training directly by converting user`s trained model.
Initial Apache Arrow support. New type of Numeric Tables were added to DAAL.
The performance was improved for the following algorithms:
Decision Forest Classification prediction for the processors supporting Intel® Advanced Vector Extensions 2 (Intel® AVX2) and Intel® Advanced Vector Extensions 512 (Intel® AVX-512)
Gradient Boosted Trees algorithm(prediction stage and model size)