Hot on the heels of Google with TensorFlow, IBM has announced that SystemML will be made freely available through Apache. SystemML was designed with the main goal of lowering the complexity required to maintain and scale machine learning algorithms. It provides a declarative machine learning (DML) language that simplifies the specification of machine learning algorithms using an R-like syntax that significantly increases the productivity of data scientist, as it provides flexibility on how the custom analytics are expressed and provides data independence from the underlying input formats and physical custom analytics.
Currently, SystemML supports processing analytic computations in a single-node in-memory mode as well as in multi-node distributed mode using Hadoop and Spark.
SystemML advances machine learning in two very important ways. The SystemML language, Declarative Machine Learning (DML), includes linear algebra primitives, statistical functions, and ML-specific constructs that make it easier and more natural to express ML algorithms. Algorithms can be expressed in either an R-like or a Python-like syntax. DML significantly increases the productivity of data scientists by providing full flexibility in expressing custom analytics as well as data independence from the underlying input formats and physical data representations.
Second, SystemML provides automatic optimization according to data and cluster characteristics to ensure both efficiency and scalability. SystemML runs in MapReduce or Spark environments.