GitHub user feynmanliang opened a pull request:
https://github.com/apache/spark/pull/8648
[SPARK-10478][ML] Performance, organization, and style improvements for
multi-layer perceptron
* Changes manual iterations into higher-performance `UFunc`s, vectorized
operations, and broadcasted operations
* Refactors multiple `while` and `for` loops into `foreach` and `map`s
* Fixes various style issues
* Adds comments and improves scaladocs
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/feynmanliang/spark ann-improvements
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/8648.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #8648
----
commit 611c76f723405fe40ed1ba5c871d5e88b617aed3
Author: Feynman Liang <[email protected]>
Date: 2015-09-07T20:00:32Z
Documentation and indentation fixes
commit 7b192db010ff660da39404009ce883f903f9980a
Author: Feynman Liang <[email protected]>
Date: 2015-09-07T20:02:36Z
Refactors unneeded helpers
commit 12169d7236caaad0db57d941cb9999f47a947b61
Author: Feynman Liang <[email protected]>
Date: 2015-09-07T20:08:05Z
More doc and style fixes
commit bc52b652541f2d6c4508662af937e0a26e207651
Author: Feynman Liang <[email protected]>
Date: 2015-09-07T20:39:21Z
Cleans up documentation and uses functional code
commit 84f8bea0f4a3b24f8aae6eadd98d7166311f951d
Author: Feynman Liang <[email protected]>
Date: 2015-09-07T21:27:53Z
Vectorizes linalg using ufuncs and vector ops
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
