It’s been a long time in the making, with lots of structural changes underneath the hood. But… A new years release is now available: 20.1.0
ADAMS website:
https://adams.cms.waikato.ac. nz/download/20-1-0/
Or directly from sourceforge.net:
https://sourceforge.net/ projects/theadamsflow/files/ 20.1.0/
https://adams.cms.waikato.ac.
Or directly from sourceforge.net:
https://sourceforge.net/
Some highlights:
- Over 70 new actors and over 20 new conversions
- Integration with RabbitMQ message broker for distribution compute jobs and sub-flows
- Communication with Python processes via Pyro4, eg for integrating scikit-learn models as regular Weka classifiers
- Better support for Java collections
- MOA updated to latest release
- Improved support for merging spreadsheets and Weka datasets
- Slimmed down adams-core module, by moving out mathematical, JSON, YAML, XML functionality into separate modules. JDBC functionality other than MySQL got moved into the adams-db module.
- Split adams-rats into separate modules: core, net, rest, webservice (SOAP)
- Moved out HTML functionality from adams-net into new module adams-html
- New adams-groovy-rest module for writing REST webservices entirely in Groovy, avoiding compilation
- Rest webservices work over https as well
- Compression/decompression actors for bzip2, gzip, lzf, lzma, xz and zstd can process raw byte arrays in addition to files
- The Weka Investigator can generate outputs also on a per-split basis when using cross-validation as evaluation method, it can also distribute cross-validation jobs via RabbitMQ to a compute farm
- Support for Debian and RPM packages, simplifying generation of docker images