The Apache MADlib team is pleased to announce the immediate
availability of MADlib version 1.16.
Highlights of the MADlib 1.16 release:
New features:
- Deep learning: support for Keras with TensorFlow backend with GPU
- acceleration
- Deep learning: utility to load model architectures and weights
- Deep learning: preprocess images for gradient descent optimization
algorithms
- kd-tree method for k-nearest neighbors for faster approximate
solution
- Support for Greenplum 6
- Support for PostgreSQL 11
Bug fixes, including:
- Jaccard distance was not releasing memory
- MLP with minibatching on postgres fixed
- MLP was not stopping after tolerance was reached
- MLP warm start fixed
- MLP with minibatch for integer dependent variable on PostgreSQL
- Pivot: array_agg/distinct scaling issue on gpdb fixed
- linregr_train with dependent variable a JSONB element fixed
- MADLib 1.15 was not recognizing Postgres 10 declarative partitioned
table
- Encoding module with bigint fixed
- SVM class_weight param fixed
Other:
- Simplified maintenance by removing online examples from sql functions
- Improved performance for weakly connected components
- SVD and other messaging improvements
- max itemset size default in assoc rules changed to 10
The complete release notes can be found here:
https://cwiki.apache.org/confluence/display/MADLIB/MADlib+1.16
You can download the source release and convenience binary packages
from Apache MADlib's download page here:
http://madlib.apache.org/download.html
Alternatively, you can download through an ASF mirror near you:
https://www.apache.org/dyn/closer.lua/madlib/1.16
----
Apache MADlib is an open-source library for scalable in-database
analytics. It provides data-parallel implementations of mathematical,
statistical and machine learning methods for structured and
unstructured data.
The MADlib mission: to foster widespread development of scalable
analytic skills, by harnessing efforts from commercial practice,
academic research, and open-source development.
We welcome your help and feedback. For more information on how to
report problems, and to get involved, visit the project website at
https://madlib.apache.org
----
Thank you, everyone who contributed to the MADlib 1.16 release. We
look forward to continued community participation for the next
release.
Warm Regards,
Domino Valdano
