Developer Guide for Intel® Data Analytics Acceleration Library 2016 Update 4
Training and prediction algorithms in Intel® Data Analytics Acceleration Library (Intel® DAAL) include a range of popular machine learning algorithms. Unlike analysis algorithms, which are intended to characterize the structure of data sets, machine learning algorithms model the data. Modeling operates in two major stages:
Training.
At this stage, the algorithm estimates model parameters based on a training data set.
Prediction or decision making.
At this stage, the algorithm uses the trained model to predict the outcome based on new data.
The following major categories of training and prediction methods are available in Intel DAAL:
Regression methods.
These methods predict the values of dependent variables (responses) by observing independent variables.
Classification methods.
These methods identify to which sub-population (class) a given observation belongs.
Recommendation methods.
These methods predict the preference that a user would give to a certain item.
Training is typically a lot more computationally complex problem than prediction. Therefore, certain end-to-end analytics usage scenarios require that training and prediction phases are done on distinct devices, the training is done on more powerful devices, while prediction is done on smaller devices. Because smaller devices may have stricter memory footprint requirements, Intel DAAL separates Training, Prediction, and respective Model in three different class hierarchies to minimize the footprint.