:orphan:

|rss_image|  **deep learning - 1/1** :ref:`Blog <ap-main-0>` :ref:`article (8) <ap-cat-article-0>` :ref:`articles (3) <ap-cat-articles-0>` :ref:`cours (6) <ap-cat-cours-0>` :ref:`module (7) <ap-cat-module-0>` :ref:`paper (3) <ap-cat-paper-0>`

.. |rss_image| image:: feed-icon-16x16.png
    :target: ../_downloads/rss.xml
    :alt: RSS

----


.. index:: deep learning


.. _ap-cat-deep_learning-0:

deep learning - 1/1
+++++++++++++++++++

.. blogpostagg::
    :title: Deep learning dans un autre projet
    :date: 2018-07-02
    :keywords: ensae_teaching_dl
    :categories: deep learning
    :rawfile: 2018/2018-07-02_split.rst

    Les notebooks de deep learning devenaient difficiles
    à maintenir dans le même projet, les librairies
    dépassent souvent la centaine de mégaoctets
    et allongeaient considérablement le temps d'exécution
    des tests unitaires. Ils ont désormais leur
    propre projet :epkg:`ensae_teaching_dl`.




.. blogpostagg::
    :title: Comparaison de librairies de deep learning
    :date: 2018-03-19
    :keywords: benchmark,pytorck,tensorflow,cntk
    :categories: deep learning
    :rawfile: 2018/2018-03-19_deep_learning_framework.rst

    C'est un article de blog produit par la recherche de Microsoft :
    `Comparing Deep Learning Frameworks: A Rosetta Stone Approach <https://blogs.technet.microsoft.com/machinelearning/2018/03/14/comparing-deep-learning-frameworks-a-rosetta-stone-approach/>`_.
    Il compare les temps d'apprentissage et de test pour la plupart
    des librairies de deep learning.
    L'ensemble est disponible sur
    `GitHub <https://github.com/ilkarman/DeepLearningFrameworks>`_.
    Le répertoire de
    `notebooks <https://github.com/ilkarman/DeepLearningFrameworks/tree/master/notebooks>`_
    est intéressant car cela permet de récupérer les mêmes exemples (4)
    implémentés avec les différantes librairies de deep learning
    les plus connues.




.. blogpostagg::
    :title: Saves and reload a pytorch model
    :date: 2018-01-07
    :keywords: torch,serialize
    :categories: deep learning
    :rawfile: 2018/2018-01-07_torch.rst

    Saving and loading a :epkg:`pytorch` model is something
    which took me a couple of searches on internet so
    let's write it out. So to save a model:
    

    ...




.. blogpostagg::
    :title: Deep learning et a priori
    :date: 2016-03-23
    :keywords: Bengio,apprentissage
    :categories: deep learning
    :rawfile: 2016/2016-03-23_deep_leanring.rst

    Le deep learning n'a plus à prouver son efficacité. L'article suivant décrit quelques cas
    pratiques et montre que la façon de mener l'apprentissage à un impact sur la
    performance finale :
    `Knowledge Matters: Importance of Prior Information for Optimization <http://jmlr.org/papers/volume17/gulchere16a/gulchere16a.pdf>`_.




.. blogpostagg::
    :title: Deep learning and other readings
    :date: 2015-12-22
    :keywords: machine learning,dask,OSM,Open Street Map
    :categories: deep learning
    :rawfile: 2015/2015-12-22_deeplearning.rst

    I came accross the following article
    `Evaluation of Deep Learning Toolkits <https://github.com/zer0n/deepframeworks/blob/master/README.md>`_
    which studies a short list of libraries for deep learning: Caffe, CNTK, TensorFlow, Theano, Torch,
    and various angles: modeling capability, interfaces, model deployment, performance, architecture, ecosystem, cross-platform.
    It gives a nice overview and helps choosing the library which fits your needs. Once your deep models has been trained,
    how to use it? This question should be the first one to be answered.
    

    ...





----

|rss_image|  **deep learning - 1/1** :ref:`2022-10 (1) <ap-month-2022-10-0>` :ref:`2022-12 (2) <ap-month-2022-12-0>` :ref:`2023-01 (1) <ap-month-2023-01-0>` :ref:`2023-02 (1) <ap-month-2023-02-0>` :ref:`2023-04 (1) <ap-month-2023-04-0>`