GithubHelp home page GithubHelp logo

davidgengenbach / openimages-fastxml-classification Goto Github PK

View Code? Open in Web Editor NEW
4.0 4.0 0.0 83.1 MB

Training a multi-label FastXML classifier on the OpenImages dataset

License: GNU General Public License v3.0

Shell 2.03% Python 20.75% Makefile 0.06% R 0.15% C++ 4.59% MATLAB 1.10% Batchfile 0.19% C 0.01% Perl 0.06% Jupyter Notebook 71.07%
classification fast-xml fastxml-classifier machine-learning multilabel-classification transfer-learning vgg16

openimages-fastxml-classification's People

Contributors

davidgengenbach avatar

Stargazers

 avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar

openimages-fastxml-classification's Issues

Wie ist Anordnung von Labelqualität?

Anordnung der Labelqualität:

Durch-Menschen-bestätigt > Durch-Menschen-verworfen > Nicht-vorhanden

oder

Durch-Menschen-bestätigt > Nicht-vorhanden > Durch-Menschen-verworfen

Agenda für nächsten Termin

  • Ergebnisse von Feature Extraction variieren
    • Zwischen einzelnen Extrahierungsimplementierungen
  • Output vom Feature-Extrahieren
    • 4096 fully connected (vorletzer Layer, ohne ReLu)
    • 1000 fully connected (letzer Layer, ohne ReLu, ohne Softmax)
  • FastXML
    • CPP
      • Trainieren braucht viel RAM (bei einem Thread ~25GB)
      • LogLoss ist sehr hoch!
      • Metriken berechnen geht nur in Batches
      • Model recht groß (3,7GB bei 10 Bäumen)
  • HOMER

DUMP

dark knowledge

Offene Fragen
Wie wurden das System für die Annotation trainiert?
Wie enstehen die Unterschiede zwischen den Annotations-per-image von Validation und Training Set?
Wieso gibt es weniger Annotations bei Humans als bei Machine?
Heiko fragen wegen Slurm-Head ("Could not chdir to home directory /nfs/home/dgengenbach: No such file or directory")!
Sind die "falschen" Labels einfach sehr gut trainierbar? Ambiguity bei mispredicted

08.01.
Homer zum Lernen anwenden (mulan)
calibrated label ranking (mlc)
positive negativ labeling
probabilistic label trees

Protokoll: Treffen 24.01.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.