Apache Spark itself

1. 

AMPLab

Spark originally came out of Berkeley AMPLab and even today AMPLab projects, even though they are not in Apache Spark Foundation, enjoy a status a bit over your everyday github project.

Spark's own MLLib forms the bottom layer of the three-layer ML Base, with MLI being the middle layer and ML Optimizer being the most abstract layer.

2. 

3. ML Optimizer (aka )

Ghostware was described in 2014 but never released. Of the 39 machine learning libraries, this is the only one that is vaporware, and is included only due to its AMPLab and ML Base status.

Other than ML Base

4. 

A recent project from June, 2015, this set of stochastic learning algorithms claims 25x - 75x faster performance than Spark MLlib on Stochastic Gradient Descent (SGD). Plus it's an AMPLab project that begins with the letters "sp", so it's worth watching.

5. 

Brought machine learning pipelines to Spark, but pipelines have matured in recent versions of Spark. Also promises some computer vision capability, but there are  I previously blogged about.

6. 

A server to manage a large collection of machine learning models.

7. 

Faster machine learning on Spark by optimizing communication patterns and shuffles, as described in the paper

Frameworks

GPU-based

8. 

 

I previously blogged 

9. 

Brand new and frankly why I started this list for this blog post. Provides an interface to .

Non-GPU-based

10. 

Parameter server for model-parallel rather than data-parallel (as Spark's MLlib is).

11. 

From Airbnb, used in their automated pricing

12. 

Logistic regression, LDA, Factorization machines, Neural Network, Restricted Boltzmann Machines

13. 

Similar to Spark DataFrames, but agnostic to engine (i.e. will run on engines other than Spark in the future). Includes cross-validation and interfaces to external machine learning libraries.

Interfaces to other Machine Learning systems

14. 

Wraps Stanford .

15. 

Interface to Python's 

16. 

Interface to 

17. 

Wraps , machine learning in Hive

18. 

Export PMML, an industry standard XML format for transporting machine learning models.

Add-ons that enhance MLlib's existing algorithms

19. 

Adds dropout capability to Spark MLLib, based on the paper .

20. 

Adds arbitrary distance functions to K-Means

21. 

Visualize the Streaming Machine Learning algorithms built into Spark MLlib

Algorithms

Supervised learning

22. 

Factorization Machines

23. 

Recursive Neural Networks (RNNs)

24. 

SVM based on the performant Spark communication framework CoCoA listed above.

25. 

Based on 

26. 

Matrix Factorization Recommendation System

Unsupervised learning

27. 

40x faster clustering than Spark MLlib K-Means

28. 

K-Means that produces more uniformly-sized clusters, based on 

29. 

Build graphs using k-nearest-neighbors and locality sensitive hashing (LSH)

30. 

Online Latent Dirichlet Allocation (LDA), Gibbs Sampling LDA, Online Hierarchical Dirichlet Process (HDP)

Algorithm building blocks

31. 

Adaboost and MP-Boost

32. 

Port to Spark of . If your machine learning cost function happens to be convex, then TFOCS can solve it.

33. 

Linear algebra operators to work with Spark MLlib's linalg package

Feature extractors

34. 

Information-theoretic basis for feature selection, based on 

35. 

Given labeled data, "discretize" one of the continuous numeric dimensions such that each bin is relatively homogenous in terms of data classes. This is a foundational idea CART and ID3 algorithms to generate decision trees. Based on .

36. 

Distributed  for dimensionality reduction.

37. 

Sparse feature vectors

Domain-specific

38. 

K-Means, Regression, and Statistics

39. 

来自：http://datascienceassn.org/content/39-machine-learning-libraries-spark-categorized