ND4S
| Original author(s) | Adam Gibson, Taisuke Oe |
|---|---|
| Developer(s) | various |
| Stable release |
0.0.3.5.5
/ 15 September 2014 |
| Development status | active |
| Written in | Java |
| Operating system | Linux, OSX, Windows, Android |
| Platform | Cross-platform |
| Type | Natural language processing, Deep learning, Machine vision |
| License | Apache 2.0 |
| Website |
github |
ND4S is a free, open-source extension of the Scala programming language operating on the Java Virtual Machine—though it is compatible with both Java and Clojure.[1]
ND4S is a scientific computing library for linear algebra and matrix manipulation in a production environment, integrating with Hadoop and Spark to work with distributed GPUs. It supports n-dimensional arrays for JVM-based languages.
ND4S has primarily been developed by the group in San Francisco that built Deeplearning4j, led by Adam Gibson.[2] It was created under an Apache Software Foundation license.
Distributed
ND4S's operations include distributed parallel versions. They can take place in a cluster and process massive amounts of data. Matrix manipulation occurs in parallel on GPUs or CPUs in the cloud, and can work within Spark or Hadoop clusters.
Benchmarks
Benchmarks show that ND4S runs roughly twice as fast as Numpy on large matrices.[3]