Skip to content

Instantly share code, notes, and snippets.

@kmaehashi
Last active October 31, 2016 04:41
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save kmaehashi/94a2708d333dcc841d490bb636afa556 to your computer and use it in GitHub Desktop.
Save kmaehashi/94a2708d333dcc841d490bb636afa556 to your computer and use it in GitHub Desktop.
Jubatus 1.0 release note candidate

Release 0.4.0 (2016-10-31)

  • New Features
    • Add Recommender service (#52, #58)
    • Add model file manipulation tools (#4, #62)
    • Add ConcatLoader (#61)
  • Improvements
    • Support Jubatus 1.0 (#66)
    • Improve Classifier Schema to accept columns without LABEL (#40, #57)
    • Improve handlign of bool-typed values (#9, #63)
    • Add option to apply Softmax to the resulting score in Classifier (#59)
    • Implement feature to enable logging by environment variable (#39, #65)
    • Improve RPC port allocation method (#44, #54)
    • Improve jubakit to work with non-standard Homebrew installations on macOS (#32, #35)
    • Update theme of docs (#60)
  • Bug Fixes
    • Fix non-static Datasets does not allow automatic Schema prediction (#41, #64)
    • Fix error handling in jubash command (#55)
    • Fix error message when Jubatus servers are not installed on macOS (#56)

Release 1.0.0 - 2016/10/31

  • New Features
    • Add image extraction plug-in (#843, #1150)
    • Add configuration validation feature (#973, #1144)
  • Improvements
    • clustering: change APIs to accept ID-based datum (#1155)
    • Update Dockerfile to use Ubuntu 14.04 (#1146)
    • Improve --daemon option of server and proxy to detach from console (#1072, #1148)
    • classifier: models are now compatible in standalone and distributed mode (#469, #1153)
    • Change the default log level to INFO instead of DEBUG (#1152)
    • Update example configurations (#1165)
    • Update manpages (#1167)
    • Update packaging tools (#1157)
  • Bug fixes
    • graph: fix IDs being reused when loading model file to other server instance in standalone mode (#1149, #1156)

Release 1.0.0 - 2016/10/31

  • New Machine Learning Algorithms
    • clustering: add DBSCAN algorithm (#330, #332)
    • regression: add perceptron, PA, PA2, CW, AROW, NHERD, NN, cosine and euclidean (#327, #328)
  • Improvements
    • clustering: improve APIs to make each Datum distinguishable by user-specified ID (#317, #326)
    • clustering: improve config format and parameter range checking (#314, #316, #325, #334, #335)
    • graph: add method to return the largest ID (#333)
    • classifier and regression: Rename diff_size to num_features_diff status key (#336)
    • Improve binary feature key names to include the rule name (#329)
    • regression: remove unused argument from driver constructor (#319, #320)
    • Remove unused revert function (#315, #331)
  • Bug fixes
    • recommender: fix rows not cleared occasionally in inverted_index and inverted_index_euclid (#318)

その他のトピックス

  • Dockerhubイメージを公開。docker pullでJubatusを簡単に動作させられるようになった。
  • Jubatus公式サイトを見やすく再構成。
  • embedded-jubatus-core機能の提供。jubatus_coreの機能をpythonから呼び出せるようになった。jubatus_coreがpythonクライアントからのシームレスな呼び出しも可能.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment