{"id":182,"date":"2018-07-26T07:57:06","date_gmt":"2018-07-26T05:57:06","guid":{"rendered":"https:\/\/user.phil.hhu.de\/bladier\/?page_id=182"},"modified":"2019-01-29T09:57:52","modified_gmt":"2019-01-29T08:57:52","slug":"deep_learning_nlp","status":"publish","type":"page","link":"https:\/\/user.phil.hhu.de\/bladier\/deep_learning_nlp\/","title":{"rendered":"Deep Learning for NLP \u2014 Kurswebseite"},"content":{"rendered":"<div id=\"pl-182\"  class=\"panel-layout\" ><div id=\"pg-182-0\"  class=\"panel-grid panel-no-style\" ><div id=\"pgc-182-0-0\"  class=\"panel-grid-cell\" ><div id=\"panel-182-0-0-0\" class=\"so-panel widget widget_sow-editor panel-first-child panel-last-child\" data-index=\"0\" ><div\n\t\t\t\n\t\t\tclass=\"so-widget-sow-editor so-widget-sow-editor-base\"\n\t\t\t\n\t\t>\n<div class=\"siteorigin-widget-tinymce textwidget\">\n\t<p><!-- Bootstrap CSS --><\/p>\n<p><!-- Our CSS --><\/p>\n<h5><span style=\"font-size: 36.75px\">Deep Learning for NLP \u2014 Kurswebseite<br \/>\n<\/span><em style=\"font-size: 20.75px\"><a href=\"https:\/\/user.phil.uni-duesseldorf.de\/~cwurm\/\">Christian Wurm<\/a> &amp; <a href=\"https:\/\/user.phil.hhu.de\/bladier\/\">Tatiana Bladier<\/a><\/em><\/h5>\n<div class=\"container\">\n<div class=\"row\">\n<div class=\"col\">\n<p>Aufbauseminar, Universit\u00e4t D\u00fcsseldorf, WiSe 2018<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div><\/div><\/div><\/div><div id=\"pg-182-1\"  class=\"panel-grid panel-no-style\" ><div id=\"pgc-182-1-0\"  class=\"panel-grid-cell\" ><div id=\"panel-182-1-0-0\" class=\"widget_text so-panel widget widget_custom_html panel-first-child panel-last-child\" data-index=\"1\" ><div class=\"textwidget custom-html-widget\"><h3> Organisatorisches <\/h3>\n\nMontag 14:30 \u2013 16:00, Raum 24.21.05.61\n<br \/>\nDienstag 10:30 \u2013 12:00, Raum 24.21.03.62-64 \n<br \/>\n<br \/>\n<b>***NEU!*** Tutorium zum Kurs: <\/b> \n<br \/>\nMontag 10:30 \u2013 12:00, Raum 24.21.03.62-64 (Tutor: Alexander Teusz)\n\n<br \/>\n<br \/>\n\n<b>Emails:<\/b> cwurm[at]phil.hhu.de, bladier[at]phil.hhu.de, Alexander[dot]Teusz[at]uni-duesseldorf.de\n<br \/>\n<b>Sprechstunde:<\/b> nach Vereinbarung\n<br \/>\n<br \/>\nZiel dieses Kurses ist es, die state-of-the-art Techniken neuronaler Netze einerseits zu verstehen, andererseits praktisch zu implementieren. Kursinhalte (zu Theorie und Praxis) werden unten angegeben. Training und Implementierung neuronaler Netze wird mit Python und Keras (einer Python-Bibliothek) umgesetzt werden. Montags werden \u2013 tendenziell \u2013 eher theoretische Grundlagen besprochen, Dienstags wird \u2013 tendenziell \u2013 programmiert.\n\n<h3> Theoretische Inhalte (<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/12\/deep_architectures.pdf\">Skript<\/a>) <\/h3>\n\nDie theoretischen Inhalte finden sich in diesem <a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/12\/deep_architectures.pdf\">Skript<\/a> (Vorsicht, \u00e4ndert sich \u00f6fters mal! Letzte Aktualisierung: 10. Dezember 2018).\n\n<h3> Beteiligungsnachweise und AP-Scheine<\/h3>\n\nEin Beteiligungsnachweis (BN) kann durch die Bearbeitung der Hausaufgaben erworben werden. F\u00fcr einen AP-Schein muss eine Hausarbeit geschrieben werden (4-5 Seiten f\u00fcr Bachelor-Studierende, 7-10 Seiten f\u00fcr Master-Studierende).\n\n<br \/>\n<br \/>\n\n[<a href=\"https:\/\/www.dropbox.com\/s\/2gysdp6brc3iixk\/presentation_template_latex.zip?dl=0\"><b><u>LaTex-Vorlage f\u00fcr die Pr\u00e4sentationen (zip-Datei)<\/u><\/b><\/a>] \n<br \/>\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/6jpnblf4x5x12rq\/HA_Backpropagation.zip?dl=0\"><b><u>LaTex-Datei f\u00fcr die Backpropagation-Hausaufgabe<\/u><\/b><\/a>] \n<br \/>\n<br \/>\nOnline LaTex-Vorlagen: <a href=\"https:\/\/www.overleaf.com\/latex\/templates\">Overleaf-Templates<\/a>\n\n<h3> Abschlussprojekte: Vorlagen f\u00fcr Projektberichte<\/h3>\n<br \/>\nEin Abschlussprojekt umfasst die Durchf\u00fchrung eines Experiments mit Neuronalen Netzen zu NLP-Fragen und das Verfassen eines kurzen Projektberichts (4-5 Seiten f\u00fcr Bachelor-Studierende, 7-10 Seiten f\u00fcr Master-Studierende). Der Projektbericht soll nach dem klassischen <a href=\"https:\/\/en.wikipedia.org\/wiki\/IMRAD\">IMRAD-Schema<\/a> mit Benutzung der unten angegebenen Vorlagen (LaTex oder Word) geschrieben werden. Der Code und die Daten k\u00f6nnen entweder als .zip-Datei abgegeben werden oder als Link zu einer (privaten oder \u00f6ffentlichen) GitHub-Repository. Mein Profil auf GitHub: <a href=\"https:\/\/github.com\/TaniaBladier\">https:\/\/github.com\/TaniaBladier<\/a>.\n<br \/>\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/azn85zj5kxog5my\/acl18-latex.zip?dl=0\"><b><u>LaTex-Vorlage f\u00fcr die Projektberichte (zip-Datei)<\/u><\/b><\/a>] \n\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/150kaxj14huhilc\/acl18-word.zip?dl=0\"><b><u>Word-Vorlage f\u00fcr die Projektberichte (zip-Datei)<\/u><\/b><\/a>] \n<br \/>\n<br \/>\n<a href=\"https:\/\/www.overleaf.com\/\"><b><u>Online LaTex Editor: Overleaf<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil-fak.uni-duesseldorf.de\/~samih\/drawing\/\"><b><u>Draw.io: Tool for drawing Neural Network Models<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2019\/01\/Nuhn_2017_Tutorial_Latex.pdf\"><b><u>Patrick Nuhn (2017) LaTex Tutorial: Getting Started.<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2019\/01\/Attia_2018_Neural_Sent_Analysis.pdf\"><b>Beispiel eines Projektberichts (Artikel von Attia et al. (2018) zu Neural Sentiment Analysis mit Tweets)<\/b><\/a>\n\n\n<h3> Hilfreiche Links und Tutorials <\/h3>\n\n<ul style=\"list-style-type:square\"> \n<li>Jason Brownlee <a href=\"https:\/\/machinelearningmastery.com\/\">\"Machine Learning Mastery\"<\/a><\/li>\n<li> Robbie Allen (2017) <a href=\"https:\/\/medium.com\/machine-learning-in-practice\/over-150-of-the-best-machine-learning-nlp-and-python-tutorials-ive-found-ffce2939bd78\" \/> \"Over 150 of the Best Machine Learning, NLP, and Python Tutorials I\u2019ve Found\"<\/a><\/li>\n<li><a href=\"https:\/\/cs231n.github.io\/python-numpy-tutorial\/\" \/> Python\/Numpy tutorial <\/a><\/li>\n<li><a href=\"https:\/\/cs231n.github.io\/ipython-tutorial\" \/> IPython\/Jupyter notebook tutorial <\/a><\/li>\n<li><a href=\"https:\/\/becominghuman.ai\/cheat-sheets-for-ai-neural-networks-machine-learning-deep-learning-big-data-678c51b4b463\" \/> Cheat Sheets for Machine Learning and Deep Learning<\/a><\/li>\n<\/ul>\n<\/div><\/div><\/div><\/div><div id=\"pg-182-2\"  class=\"panel-grid panel-no-style\" ><div id=\"pgc-182-2-0\"  class=\"panel-grid-cell\" ><div id=\"panel-182-2-0-0\" class=\"widget_text so-panel widget widget_custom_html panel-first-child panel-last-child\" data-index=\"2\" ><div class=\"textwidget custom-html-widget\"><hr class=\"style-two\">\n\n\n<h2> Kursinhalte <\/h2>\n\n<br \/>\n<br \/>\n\n[<a href=\"http:\/\/etherpad.phil-fak.uni-duesseldorf.de\/iGEH5BfJg1\"><b><u>Link zum EtherPad<\/u><\/b><\/a>]<\/div><\/div><\/div><\/div><div id=\"pg-182-3\"  class=\"panel-grid panel-has-style\" ><div id=\"start\" class=\"panel-row-style panel-row-style-for-182-3\" ><div id=\"pgc-182-3-0\"  class=\"panel-grid-cell panel-grid-cell-mobile-last\" ><div id=\"panel-182-3-0-0\" class=\"widget_text so-panel widget widget_custom_html panel-first-child\" data-index=\"3\" ><div id=\"w1\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-0\" ><h3 class=\"widget-title\">Woche 1: Einf\u00fchrung und Semester\u00fcberblick<\/h3><div class=\"textwidget custom-html-widget\">08.10.2018 Semester\u00fcberblick, <a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/10\/01_DL_Intro.pdf\"><u>Folien zum praktischen Teil<\/u><\/a>.\n<br \/>\n09.10.2018 Theoretische Sitzung: Methodik des maschinellen Lernens\n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> Goldberg, Y. (2016). <a href=\"http:\/\/u.cs.biu.ac.il\/~yogo\/nnlp.pdf\"> A primer on neural network models for natural language processing.<\/a> <br \/> Journal of Artificial Intelligence Research, 57, 345-420.<\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-1\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"4\" ><div id=\"w2\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-1\" ><h3 class=\"widget-title\">Woche 2: Lineare Algebra \/ Boston Housing Project<\/h3><div class=\"textwidget custom-html-widget\">15.10.2018 Wahrscheinlichkeitstheorie + wichtige Begriffe.\n<br \/>\n16.10.2018 <a href=\"https:\/\/www.dropbox.com\/s\/vthdilktjklmqqs\/boston_housing_exercise.zip?dl=0\"> <b><u>Praktische Sitzung: Boston Housing Project.<\/u><\/b><\/a> \n<br \/>\n<a href=\"http:\/\/playground.tensorflow.org\/#activation=tanh&amp;batchSize=10&amp;dataset=circle&amp;regDataset=reg-plane&amp;learningRate=0.03&amp;regularizationRate=0&amp;noise=30&amp;networkShape=5&amp;seed=0.37743&amp;showTestData=false&amp;discretize=false&amp;percTrainData=80&amp;x=true&amp;y=true&amp;xTimesY=false&amp;xSquared=false&amp;ySquared=false&amp;cosX=false&amp;sinX=false&amp;cosY=false&amp;sinY=false&amp;collectStats=false&amp;problem=regression&amp;initZero=false&amp;hideText=false\"> <b><u>Google Tool to Tinker With a Neural Network.<\/u><\/b><\/a> \n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/10\/02_DL_NLP_Project_Inspirations.pdf\"><u>Deep Learning in NLP: Project Inspirations<\/u><\/a> \n<br \/>\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/10\/Homework_Boston_Housing.pdf\"> <b><u>Hausaufgabe bis 23. Oktober, 10:00<\/u><\/b><\/a>  \n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/f9sqpgzocai87e1\/Homework_Boston_Housing_Solution.zip?dl=0\"> <b><u>Boston Housing Aufgabe: L\u00f6sung<\/u><\/b><\/a>]<\/div><\/div><\/div><div id=\"panel-182-3-0-2\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"5\" ><div id=\"w3\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-2\" ><h3 class=\"widget-title\">Woche 3: Character-level RNN for name classification (1)<\/h3><div class=\"textwidget custom-html-widget\">22.10.2018 Vektoren und Matrizen, generalisierte lineare Modelle.\n<br \/>\n23.10.2018 Praktische Sitzung: Character-level Recurrent NN for name recognition.  \n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/edz2m711jxt9ql1\/181030_2_class_version_simple_RNN.zip?dl=0\"><b><u>Code und Daten, updated 30 October<\/u><\/b><\/a>]\n\n\n<br \/>\n<br \/>\n\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/10\/ha1.pdf\"><b><u>Theoretische Hausaufgabe (bis Dienstag, 30. Oktober, 10:30, pers\u00f6nlich oder per E-Mail)<\/u><\/b><\/a>\n\n<br \/>\n<br \/>\n\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"http:\/\/colah.github.io\/posts\/2015-08-Understanding-LSTMs\/\"><u>Tutorial zu RNNs<\/u><\/a><\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-3\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"6\" ><div id=\"w4\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-3\" ><h3 class=\"widget-title\">Woche 4: Character-level RNN for name classification (2)<\/h3><div class=\"textwidget custom-html-widget\">29.10.2018 Nichtlineare Funktionen auf Vektoren.\n<br \/>\n30.10.2018 Praktische Sitzung: Character-level RNN for name classification [<a href=\"https:\/\/www.dropbox.com\/s\/edz2m711jxt9ql1\/181030_2_class_version_simple_RNN.zip?dl=0\"><b><u>Code und Daten, updated 30 October <\/u><\/b><\/a>]\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/10\/Homework_Data_Pre-Processing_Neural_POS-tagging.pdf\"><b><u>Hausaufgabe (bis Dienstag, 6. November, 10:00)<\/u><\/b><\/a>\n<br \/>\n<a href=\"https:\/\/www.dropbox.com\/s\/ft1wvghszcpm9im\/Hausaufgabe_POS_Tagging_Solution.zip?dl=0\"><b><u>Hausaufgabe (L\u00f6sung)<\/u><\/b><\/a>\n<br \/>\n<a href=\"http:\/\/www.surdeanu.info\/mihai\/teaching\/ista555-fall13\/readings\/PennTreebankConstituents.html\"><b><u>Penn Treebank POS tags<\/u><\/b><\/a>, <a href=\"https:\/\/www.clips.uantwerpen.be\/conll2000\/chunking\/\"><b><u>Chunking<\/u><\/b><\/a>, <a href=\"https:\/\/www.clips.uantwerpen.be\/conll2003\/ner\/\"><b><u>Named Entity Recognition<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Literatur (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/realpython.com\/python-keras-text-classification\/\">Practical Text Classification With Python and Keras<\/a><\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-4\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"7\" ><div id=\"w5\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-4\" ><h3 class=\"widget-title\">Woche 5: Neural POS-tagging + word embeddings<\/h3><div class=\"textwidget custom-html-widget\">05.11.2018 Multilayer Perzeptronen.\n<br \/>\n06.11.2018 Praktische Sitzung: Neural POS-tagging mit Recurrent Neural Networks. [<a href=\"https:\/\/www.dropbox.com\/s\/zsss1kn0u1bv6cb\/notebook_class_version_181106.zip?dl=0\"><b><u>Code und Daten, aktualisiert am 6. November<\/u><\/b><\/a>]\n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> Jacob Zweig (2018). <a href=\"https:\/\/towardsdatascience.com\/elmo-embeddings-in-keras-with-tensorflow-hub-7eb6f0145440\"> Elmo Embeddings in Keras with TensorFlow hub.<\/a> <\/li>\n<li> Chris McCormick (2016).\t<a href=\"http:\/\/mccormickml.com\/2016\/04\/19\/word2vec-tutorial-the-skip-gram-model\/\"> Word2Vec Tutorial - The Skip-Gram Model <\/a> <\/li>\n<\/ul>\n\n<\/div><\/div><\/div><div id=\"panel-182-3-0-5\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"8\" ><div id=\"w6\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-5\" ><h3 class=\"widget-title\">Woche 6: Neural POS-tagging + word embeddings<\/h3><div class=\"textwidget custom-html-widget\">12.11.2018 Backpropagation und Gradient Descent.\n<br \/>\n13.11.2018 Praktische Sitzung: Neural POS-tagging + word embeddings (2) [<a href=\"https:\/\/www.dropbox.com\/s\/8qdw0kcp5jl9ywl\/Neural_POS_tagging_full_version.zip?dl=0\"><b><u>Code und Daten, aktualisiert am 12. November<\/u><\/b><\/a>].\n<br \/>\n<br \/>\n[<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/11\/Untitled1.pdf\"><b><u>Keras Visualisierung und Implementierung (Handout von heute)<\/u><\/b><\/a>].\n<br \/>\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/pzg6fm9pskmrjwd\/homework_keras_visualization.zip?dl=0\"><b><u>Hausaufgabe (Abgabe bis 20. November)<\/u><\/b><\/a>].\n<br \/>\n<br \/>\n\n[<a href=\"https:\/\/www.dropbox.com\/s\/cc8nayqzrgjpxrt\/homework_keras_visualization_solution.zip?dl=0\"><b><u>Hausaufgabe (L\u00f6sung)<\/u><\/b><\/a>].\n<br \/>\n<br \/>\n\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/nlp.stanford.edu\/projects\/glove\/\"> GloVe pre-trained Word Embeddings <\/a><\/li>\n\t<li> <a href=\"https:\/\/embeddings.sketchengine.co.uk\/static\/index.html\"> Sketch Engine pre-trained Word Embeddings <\/a><\/li>\n\t<li> <a href=\"https:\/\/allennlp.org\/elmo\">ELMo pre-trained Word Embeddings<\/a><\/li>\n\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-6\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"9\" ><div id=\"w7\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-6\" ><h3 class=\"widget-title\">Woche 7: Neural POS-tagging + word embeddings: Word2Vec, GloVe, Sketch Engine, ELMo<\/h3><div class=\"textwidget custom-html-widget\">19.11.2018 (TBA)\n<br \/>\n20.11.2018 Praktische Sitzung: Neural POS-tagging + word embeddings: Word2Vec, GloVe, Sketch Engine, ELMo.\n<br \/>\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/devlp3c435qdswj\/neural_pos_tagging_word_embeddings.zip?dl=0\"><b>Neural POS-tagging mit GloVe Word Embeddings, Code und Daten<\/b><\/a>]\n<br \/>\n<br \/>\n[<a href=\"https:\/\/www.dropbox.com\/s\/lcier2vn0tdnrqi\/word2vec_with_the_great_gatsby.zip?dl=0\"><b>Learning Word Embeddings mit Word2Vec, Code und Daten<\/b><\/a>]\n<br \/>\n<br \/>\n[<a href=\"http:\/\/bionlp-www.utu.fi\/wv_demo\/\"><b>Word2vec: Online Demo<\/b><\/a>]\n<br \/>\n<br \/>\n\n[<a href=\"https:\/\/ronxin.github.io\/wevi\/\"><b>Wevi: word embedding visual inspector<\/b><\/a>]\n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Literatur:<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/radimrehurek.com\/gensim\/models\/word2vec.html\">Word2vec embeddings: Tutorial<\/a><\/li>\n\t<li> <a href=\"https:\/\/nlp.stanford.edu\/projects\/glove\/\"> GloVe pre-trained Word Embeddings <\/a><\/li>\n\t<li> <a href=\"https:\/\/embeddings.sketchengine.co.uk\/static\/index.html\"> Sketch Engine pre-trained Word Embeddings <\/a><\/li>\n\t<li> <a href=\"https:\/\/allennlp.org\/elmo\">ELMo pre-trained Word Embeddings<\/a><\/li>\n<\/ul>\n\n<\/div><\/div><\/div><div id=\"panel-182-3-0-7\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"10\" ><div id=\"w8\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-7\" ><h3 class=\"widget-title\">Woche 8: Sentiment analysis for tweets (1)<\/h3><div class=\"textwidget custom-html-widget\">26.11.2018 (TBA)\n<br \/>\n27.11.2018 Praktische Sitzung: Sentiment analysis for tweets. [<a href=\"https:\/\/www.dropbox.com\/s\/oszobk248bo3pbz\/Sentiment_Analysis_Tweets_Class_Version.zip?dl=0\"><b><u>Aufgabenstellung und Daten<\/u><\/b><\/a>]\n<br \/>\n<br \/>\n<b>Word Embeddings f\u00fcr Tweets:<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/www.dropbox.com\/s\/am9v7sbps8q55vc\/glove.twitter.27B.100d.zip?dl=0\">GloVe Word Embeddings f\u00fcr Tweets [glove.twitter.27B.100d.zip]<\/a><\/li>\n<\/ul>\n<br \/>\n<br \/>\n<b>Datensatz mit Tweets:<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/www.dropbox.com\/s\/gyjycv80y616w0h\/SemEval2013_Twitter_Data.zip?dl=0\">SemEval2013_Twitter_Dataset<\/a><\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-8\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"11\" ><div id=\"w9\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-8\" ><h3 class=\"widget-title\">Woche 9: Sentiment analysis for tweets (2)<\/h3><div class=\"textwidget custom-html-widget\">03.12.2018 (TBA)\n<br \/>\n04.12.2018 Praktische Sitzung: Sentiment analysis for tweets.\n\n<br \/>\n<br \/>\n<\/div><\/div><\/div><div id=\"panel-182-3-0-9\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"12\" ><div id=\"w10\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-9\" ><h3 class=\"widget-title\">Woche 10: Sentiment analysis for tweets (3)<\/h3><div class=\"textwidget custom-html-widget\">10.12.2018 (TBA)\n<br \/>\n11.12.2018 Praktische Sitzung: Sentiment analysis for tweets.\n\n<br \/>\n<br \/>\n\n<b>Code und Daten:<\/b>\n<br \/>\n<br \/>\n<a href=\"https:\/\/www.dropbox.com\/s\/js5iptiz225l9ak\/Sentiment_Analysis_Tweets_updated.zip?dl=0\"> <b><u>[Sentiment Analysis for Tweets 11.12.18, updated]<\/u><\/b><\/a>\n<\/div><\/div><\/div><div id=\"panel-182-3-0-10\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"13\" ><div id=\"w11\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-10\" ><h3 class=\"widget-title\">Woche 11: Neural machine translation with seq2seq models<\/h3><div class=\"textwidget custom-html-widget\">17.12.2018 (TBA)\n<br \/>\n18.12.2018 Praktische Sitzung: seq2seq models for machine translation. <a href=\"https:\/\/drive.google.com\/file\/d\/0BzdJvZytWnuzdjIwSWI2Mnk1ZHc\/view\"> Hassan Sajjad and Fahim Dalvi, DGfS Fall School 2017 \"Sequence to sequence models\".<\/a>\n<br \/>\n<br \/>\n<b>Code und Daten:<\/b>\n<br \/>\n<br \/>\n<a href=\"https:\/\/www.dropbox.com\/s\/6nun6xie3l873h3\/08_Seq2seq_Machine_Translation.zip?dl=0\"> <b><u>[seq2seq neural machine translation, updated 18 December]<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2018\/12\/HA_Backpropagation.pdf\"><b><u>Hausaufgabe (bis Dienstag, 8. Januar, 10:30)<\/u><\/b><\/a>\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> Keras 10 minute introduction to seq2seq models <a>https:\/\/blog.keras.io\/a-ten-minute-introduction-to-sequence-to-sequence-learning-in-keras.html<\/a><\/li>\n\t<li> Ilya Sutskever, Oriol Vinyals, Quoc V. Le (2014). <a href=\"https:\/\/arxiv.org\/pdf\/1409.3215.pdf\"> Sequence to Sequence Learning\nwith Neural Networks.<\/a> <\/li>\n<li> Cho et al. (2014).\t<a href=\"https:\/\/arxiv.org\/pdf\/1406.1078.pdf\"> Learning Phrase Representations using RNN Encoder\u2013Decoder for Statistical Machine Translation\n <\/a> <\/li>\n\t<li> Bahdanau et al. (2015).\t<a href=\"https:\/\/arxiv.org\/pdf\/1409.0473.pdf\"> Neural Machine Translation by Jointly Learning to Align and Translate\n <\/a> <\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-11\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"14\" ><div id=\"w12\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-11\" ><h3 class=\"widget-title\">Woche 12: Neural Semantic Role Labeling (1)<\/h3><div class=\"textwidget custom-html-widget\">07.01.2019 Semantic Role Labeling. [<a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2019\/01\/09_Neural_SRL.pdf\"><b><u>Slides<\/u><\/b><\/a>]   \n<br \/>\n08.01.2019 Praktische Sitzung: End-to-end Neural Semantic Role Labeling. [<a href=\"https:\/\/www.dropbox.com\/s\/x6mr8bivwfq9khs\/srl_code_data.zip?dl=0\"><b><u>Code und Daten<\/u><\/b><\/a>]  \n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/> \n<br \/>\n<ul style=\"list-style-type:square\">\n\t<li> Jie Zhou, Wei Xu (2015). <a href=\"http:\/\/www.anthology.aclweb.org\/P\/P15\/P15-1109.pdf\"> End-to-end Learning of Semantic Role Labeling Using Recurrent Neural\nNetworks.<\/a> <\/li>\n\t<li> <a href=\"https:\/\/diegma.github.io\/slides\/TutorialNNforSRL.pdf\">Diego Marcheggiani et al. (2017) Tutorial: Neural Methods for Semantic Role Labeling<\/a><\/li>\n<\/ul><\/div><\/div><\/div><div id=\"panel-182-3-0-12\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"15\" ><div id=\"w13\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-12\" ><h3 class=\"widget-title\">Woche 13: Neural Semantic Role Labeling (2)<\/h3><div class=\"textwidget custom-html-widget\">14.01.2019 (TBA)\n<br \/>\n15.01.2019 Praktische Sitzung: Fit generator and evaluation metrics. Introduction to LaTex. [<a href=\"https:\/\/www.dropbox.com\/s\/x6mr8bivwfq9khs\/srl_code_data.zip?dl=0\"><b><u>Code und Daten, updated 15 January<\/u><\/b><\/a>].\n\n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n(tba)<\/div><\/div><\/div><div id=\"panel-182-3-0-13\" class=\"widget_text so-panel widget widget_custom_html\" data-index=\"16\" ><div id=\"w15\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-13\" ><h3 class=\"widget-title\">Woche 14: Neural text generation<\/h3><div class=\"textwidget custom-html-widget\">22.01.2019 Theoretische Sitzung: Evaluation Metrics<\/div><\/div><\/div><div id=\"panel-182-3-0-14\" class=\"widget_text so-panel widget widget_custom_html panel-last-child\" data-index=\"17\" ><div id=\"w14\" class=\"widget_text panel-widget-style panel-widget-style-for-182-3-0-14\" ><h3 class=\"widget-title\">Woche 15: Neural Text Generation<\/h3><div class=\"textwidget custom-html-widget\">28.01.2019 <u><b><a href=\"https:\/\/user.phil.hhu.de\/bladier\/wp-content\/uploads\/sites\/59\/2019\/01\/10_Course_Summary.pdf\"> Summary \/ Takeaway<\/a><\/b><\/u>\n<br \/>\n29.01.2019 Praktische Sitzung: Neural text generation. [<u><b><a href=\"https:\/\/www.dropbox.com\/s\/ju7q1qy5wpdn1ir\/Text_Generation_Code_Data.zip?dl=0\"> Code and Data<\/a><\/b><\/u>]\n\n<br \/>\n<br \/>\n<b>Zus\u00e4tzliche Lekt\u00fcre (f\u00fcr Interessierte):<\/b>\n<br \/>\n<br \/>\n\n\n<ul style=\"list-style-type:square\">\n\t<li> <a href=\"https:\/\/nanogenmo.github.io\/\"> NaNoGenMo: National Novel Generation Month. Write Code that generates a novel of 50k+ words.<\/a> <\/li>\n\t<li><a href=\"https:\/\/minimaxir.com\/2018\/05\/text-neural-networks\/\"> Max Woolf: How to quickly train your own text-generating neural network.<\/a><\/li>\n\t<li><a href=\"https:\/\/machinelearningmastery.com\/text-generation-lstm-recurrent-neural-networks-python-keras\/\"> Jason Brownlee: Text Generation With LSTM RNNs in Keras. <\/a><\/li>\n\t<li><a href=\"https:\/\/cs.stanford.edu\/~zxie\/textgen.pdf\"> Ziang Xie: Neural Text Generation: A Practical Guide. <\/a><\/li>\n\t<li><a href=\"https:\/\/medium.com\/phrasee\/neural-text-generation-generating-text-using-conditional-language-models-a37b69c7cd4b\"> Neil Yager: How to generate text using conditional language models. <\/a><\/li>\n<\/ul><\/div><\/div><\/div><\/div><div id=\"pgc-182-3-1\"  class=\"panel-grid-cell panel-grid-cell-empty\" ><\/div><\/div><\/div><\/div>","protected":false},"excerpt":{"rendered":"<p>Deep Learning for NLP \u2014 Kurswebseite Christian Wurm &amp; Tatiana Bladier Aufbauseminar, Universit\u00e4t D\u00fcsseldorf, WiSe 2018<\/p>\n <a class=\"more-link\" href=\"https:\/\/user.phil.hhu.de\/bladier\/deep_learning_nlp\/\"><span class=\"more-msg\">Continue reading &rarr;<\/span><\/a>","protected":false},"author":66,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"ngg_post_thumbnail":0},"_links":{"self":[{"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/pages\/182"}],"collection":[{"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/users\/66"}],"replies":[{"embeddable":true,"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/comments?post=182"}],"version-history":[{"count":121,"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/pages\/182\/revisions"}],"predecessor-version":[{"id":412,"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/pages\/182\/revisions\/412"}],"wp:attachment":[{"href":"https:\/\/user.phil.hhu.de\/bladier\/wp-json\/wp\/v2\/media?parent=182"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}