Skip to content

Instantly share code, notes, and snippets.

View kmaehashi's full-sized avatar
📛

Kenichi Maehashi kmaehashi

📛
135 followers · 82 following
View GitHub Profile
@kmaehashi
kmaehashi / jubakit.rst
Last active August 24, 2016 10:36
jubatus 0.9.4 changelog candidate

Release 0.3.0 (2016-08-29)

  • New Features
    • Add Shell feature and jubash command (#38, #45, #51)
  • Improvements
    • Add API reference and design documents (#33, #36, #43, #53)
    • Add from_data method to Classifier service (#46)
@kmaehashi
kmaehashi / !README.md
Last active July 27, 2016 12:36
sensorbee-join 
build_sensorbee
./sensorbee run &
./sensorbee t c test

./sensorbee shell -t test < 01_select.bql &
./sensorbee shell -t test < 02_resume.bql
@kmaehashi
kmaehashi / ml.txt
Last active July 25, 2016 07:39
ann_ml jubatus 0.9.3
Hi Jubatus guys,
We are glad to announce that Jubatus 0.9.3 is now available.
In this release we improved the "split" string_types, which extracts features by splitting strings by separator, to accept UTF-8 separator characters.
In addition, we have implemented IDF-1 in the global weighting, which provides weighting equivalent to TfidfTransformer of scikit-learn. By using idf1, you can workaround the problem that features in the first document and features that appear in all documents are not added to the model.
The full changelogs are as follows:
===============================================
@kmaehashi
kmaehashi / jubakit.md
Last active July 21, 2016 07:36
jubatus 0.9.3 changelog candidate

Release 0.2.2 (2016-07-25)

  • Improvements

    • Add hyper-parameter tuning example using hyperopt library (#28)
    • Fix warnings when using numpy record (#29)

  • Bug Fixes

    • Fix log output from server not captured correctly (#30, #31)
@kmaehashi
kmaehashi / jubakit.md
Last active June 27, 2016 00:54
jubatus 0.9.2 changelog

Release 0.2.1 (2016-06-27)

  • Improvements
    • Add support for cosine and euclidean method of Classifier (#27)
    • Improve default configuration of NN method of Classifier to use multiple CPU cores (#26)
    • Improve Anomaly service to validate invalid method name (#25)
    • Fix deprecated test warning in Python 3.5 (#24)
@kmaehashi
kmaehashi / README.md
Last active June 20, 2016 08:50
BM25 RPC benchmark
@kmaehashi
kmaehashi / jubakit.md
Last active May 27, 2016 09:41
release note candidate

Release 0.2.0 (2016-05-30)

  • New Features

    • Add support for Anomaly engine (#20)
    • Add support for Weight engine (#19)

  • Improvements

    • Add logging system (#12, #17)

  • Add get_status API to Service (#15)

@kmaehashi
kmaehashi / anomaly_lof_inverted_index_euclid_with_ignore_kth_same_point.json
Last active May 17, 2016 06:52
{
"converter" : {
"string_filter_types" : {},
"string_filter_rules" : [],
"num_filter_types" : {},
"num_filter_rules" : [],
"string_types" : {},
"string_rules" : [
{ "key" : "*", "type" : "str", "sample_weight" : "bin", "global_weight" : "bin" }
],
@kmaehashi
kmaehashi / jubatus_ml.txt
Created April 25, 2016 10:08
Hi Jubatus guys,
We are proud to announce that Jubatus 0.9.0, our 8th major update, is now available.
In this release we have a plenty of new features and improvements:
- We released a new Python package called `jubakit`, a module to easily evaluate precision of models.
See https://github.com/jubatus/jubakit for details.
- We now officially support Ubuntu 14.04 Trusty.
- Improved Nearest Neighbor engine performance up to 20x.
@kmaehashi
kmaehashi / jubatus.md
Last active April 22, 2016 08:06
jubatus 0.9.0 release note candidate

Release 0.9.0 - 2016/04/25

  • New feature

    • jubakit: Python module to easily evaluate precision of models (#1044)

  • Changes to Supported Systems

    • Ubuntu 14.04 LTS is now officially supported (#1071)

  • Improvements