April 5, 2016
MACROS 6.6 Release
DATADVANCE development team is pleased to announce the release of MACROS 6.6, a new stable version of the MACROS library and featured plug-ins.
Note: to show the tight integration of MACROS algorithms and methods into the pSeven design space exploration platform, MACROS to be rebranded as pSeven Core in one of the future versions.
Major updates in MACROS 6.6 include:
- Improved robust design optimization (RDO) methods: new RDO algorithm effectively using internal surrogate modeling (surrogate based optimization) in order to significantly reduce the required number of objective evaluations. This method was specifically designed for robust design optimization problems with computationally expensice objective functions.
- Smart training of approximation models: the SmartSelection technology automatically selects the best modeling technique and tunes its parameters in order to train the most accurate model.
- Further increased effectiveness of distributed model training: now MACROS supports training parallelization over different components of the model's output and over combinations of values of categorical variables, if they are defined.
- Less memory usage and model loading speedup: MACROS 6.6 introduces a new model storage format that allows to read and write specific parts of the model.