Free Deep Learning Book (MIT Press)

The Deep Learning textbook is a resource intended to help students and practitioners enter the field of machine learning in general and deep learning in particular. The online version of the book is now complete and will remain available online for free.

Source for picture: click here


Table of Contents



1 Introduction

Part I: Applied Math and Machine Learning Basics

  • 2 Linear Algebra
  • 3 Probability and Information Theory
  • 4 Numerical Computation
  • 5 Machine Learning Basics

Part II: Modern Practical Deep Networks

  • 6 Deep Feedforward Networks
  • 7 Regularization for Deep Learning
  • 8 Optimization for Training Deep Models
  • 9 Convolutional Networks
  • 10 Sequence Modeling: Recurrent and Recursive Nets
  • 11 Practical Methodology
  • 12 Applications

Part III: Deep Learning Research

  • 13 Linear Factor Models
  • 14 Autoencoders
  • 15 Representation Learning
  • 16 Structured Probabilistic Models for Deep Learning
  • 17 Monte Carlo Methods
  • 18 Confronting the Partition Function
  • 19 Approximate Inference
  • 20 Deep Generative Models

To access the book, click here.

Рубрика: Ресурсы/Софт | Добавить комментарий

PhD fellow in Natural Language Processing and Machine Learning

Department of Computer Science, Faculty of Science at University of Copenhagen is offering a PhD scholarship in Natural Language Processing and Machine Learning (NLP/ML) commencing 1 October 2017 or as soon as possible thereafter.

Description of the scientific environment  
The NLP group at the Department of Computer Science is led by Professor Anders Søgaard and is one of the best in the world. See current members at The Department of Computer Science also has strong research groups in related fields such as Information Retrieval, Machine Learning, and Computer Vision.
Project description  
The candidate is expected to provide a project description of about 2-4 pages. While the research group has had a focus on transfer- and multi-task learning in recent years, we welcome projects in any area of NLP.

Principal supervisor is Professor Anders Søgaard, Department of Computer Science.

Читать далее

Рубрика: Вакансии/Стажировки, Курсы/Образование/Постдоки | Добавить комментарий


Хакатон направлен на разработку и оценку алгоритмов определения плагиата на русском языке с упором на научные тексты (академический плагиат). Задача, предложенная для участников, будет аналогична заданию по выравнианию текстов (Text Alignment) с соревнований PAN (, т.е. в паре текстов, перефразированные или скопированные фрагменты взятые из одного текста нужно найти во втором тексте. Для задания организаторы представят обучающий корпус, участникам же предлагается разработать и обучить свои модели на этом корпусе.

Зарегистрироваться на хакатон можно будет в день проведения или заранее, заполнив форму:

Задать вопросы и получить дополнительную информацию можно по адресу:

Обратите внимание, что для участия в хакатоне необходимо будет оплатить регистрационный взнос как за индустриальный день.

Рубрика: Конференции, Ресурсы/Софт | Добавить комментарий

Call for Chapters: Techno-Social Systems for Modern Economical and Governmental Infrastructures

Call for Chapters

Proposals Submission Deadline: August 13, 2017
Full Chapters Due: October 2, 2017


Alexander Troussov and Sergey Maruev (The Russian Presidential Academy of National Economy and Public Administration)
Submission Date: October 29, 2017

This book covers theoretical and practical aspects of mining techno-social systems which have potential for the creation of scalable methods and applications for business, governance and economics.


Topics include but are not limited to:

Geographical and Internet sociology and demographics
Mining of large volumes of governance-based records and other tabular data
Natural Language Engineering
Recommender and advertising systems, data navigation
Semantic Web, Social Web
The use of social media to fight crimes (including narcotrafficking, internal and external banking fraud)

All published chapters will be indexed by SCOPUS.


Submission Procedure

Researchers and practitioners are invited to submit on or before August 13, 2017, a chapter proposal of 1,000 to 2,000 words clearly explaining the mission and concerns of his or her proposed chapter. Authors will be notified by August 21, 2017 about the status of their proposals and sent chapter guidelines. Full chapters are expected to be submitted by October 9, 2017, and all interested authors must consult the guidelines for manuscript submissions at to submission. All submitted chapters will be reviewed on a double-blind review basis. Contributors may also be requested to serve as reviewers for this project.

Note: There are no submission or acceptance fees for manuscripts submitted to this book publication, Techno-Social Systems for Modern Economical and Governmental Infrastructures. All manuscripts are accepted based on a double-blind peer review editorial process.

All proposals should be submitted through the eEditorial Discovery®TM online submission manager.


This book is scheduled to be published by IGI Global (formerly Idea Group Inc.), publisher of the «Information Science Reference» (formerly Idea Group Reference), «Medical Information Science Reference,» «Business Science Reference,» and «Engineering Science Reference» imprints. For additional information regarding the publisher, please visit This publication is anticipated to be released in 2018.

Important Dates

August 13, 2017: Proposal Submission Deadline
August 21, 2017: Notification of Acceptance
October 2, 2017: Full Chapter Submission
October 22, 2017: Review Results Returned
October 29, 2017: Final Acceptance Notification
November 5, 2017: Final Chapter Submission


Alexander Troussov,
The Russian Presidential Academy of National Economy and Public Administration,

Рубрика: Журналы, Книги | Добавить комментарий

The 1st Workshop on Teaching NLP for Digital Humanities (Teach4DH)

—  Deadline extension: Sunday July, 23  —

Held in conjunction with GSCL 2017 (,
Berlin, Germany, September 12, 2017


Computational linguists often teach digital humanities (DH) modules that focus on using / applying / adapting NLP technologies and resources to DH problems. The challenge in such modules is to introduce students with a background in humanities or social sciences to technical content without going into detail. NLP tools often require special input formats, knowledge of program options, and post-processing of the output. NLP resources are often not easily accessible and provide only limited interaction. Interfaces and search tools require knowledge of specialized search syntax.

The workshop is intended to provide a forum for such ?NLP teachers? to share experiences, discuss best practices, introduce teaching concepts, and present demos of existing technology. It also provides an opportunity for DH researchers to express their needs and provide directions for future DH curricular developments. The workshop is intended to foster collaborations and to cross-fertilize knowledge and approaches across DH disciplines.

Teach4DH is co-organized by GSCL’s SIG Education and Profession and supported by CLARIN.


We welcome submissions of long and short papers, posters, and demonstrations relating to any aspect of teaching NLP in DH classes, including:

— Didactic units / concepts / tools for employing NLP in teaching «on stage»
— Challenges in teaching for a non-CL audience
— Exploiting NLP tools «behind the scenes» for creating didactic DH material
— Curricular considerations in developing standards for DH programs

See the website at for more details, including submission instructions. If you have any questions, please feel free to contact the workshop co-chairs at<>.


All submission deadlines are at 11:59 p.m. PST

*** Jul 23rd, 2017 Extended deadline ***
Aug 09th, 2017: Notification of acceptance
Aug 25th, 2017: Camera-ready papers due
Sep 12th, 2017: Workshop day



Peggy Bockwinkel University of Stuttgart
Thierry Declerck DFKI, Saarbrücken
Sandra Kübler Indiana University, Bloomington
Heike Zinsmeister Universität Hamburg


Melanie Andresen Universität Hamburg
Fabian Barteld Universität Hamburg
Sabine Bartsch Technische Universität Darmstadt
Noah Bubenhofer Zurich University
Stefanie Dipper Ruhr-Universität Bochum
Kim Gerdes Sorbonne nouvelle Paris
Evelyn Gius Universität Hamburg
Fotis Jannidis Julius-Maximilians-Universität Würzburg
Jonas Kuhn University of Stuttgart
Lothar Lemnitzer BBAW, Berlin
Harald Lüngen IDS, Mannheim
Nils Reiter University of Stuttgart
Thomas Schmidt IDS, Mannheim
Ulrike Schneider Johannes-Gutenberg-Universität Mainz
Julian Schulz LMU, München
Olga Scrivner Indiana University
Caroline Sporleder Georg-August-Universität Göttingen
Jannik Strötgen MPI, Saarbrücken
Thorsten Trippel University of Tübingen
Gabriel Viehhauser University of Stuttgart
Andreas Witt University of Cologne
Amir Zeldes Georgetown University

Рубрика: Конференции | Добавить комментарий

PhD position in deep learning and NLP


A PhD position is available within an industrially supported project. The
project seeks to develop deep learning methods for predictive analysis of
complex network data (e.g., social networks, trading networks etc),
focusing on statistical modelling of information exchange in transaction
networks. This is an on-going project with 2 PIs: Ivan Titov
and Max Welling

The student will be affiliated with the University of Amsterdam, but will
be mostly located at the University of Edinburgh, hosted by the Institute
of Language, Cognition and Computation (ILCC) <>
and Edinburgh NLP group <>.
ILCC is the largest lab focusing on NLP in Europe and one of the largest in
the world, with over 10 faculty doing research in this area. This PhD
vacancy will focus primarily on deep learning for natural language
processing (supervised by Ivan Titov)

We are exploiting data provided by an industrial partner. As information in
these networks mostly comes in a textual form, we are developing methods
for inducing predictive semantics representations of texts, relying both on
the text itself but also on the flow of information in the network.


University of Edinburgh

Educational level:

Master Degree

How to apply:

About The University of Edinburgh

The University’s mission is the creation, dissemination and curation of knowledge.

As a world-leading centre of academic excellence we aim to:

Enhance our position as one of the world’s leading research and teaching universities and to measure our performance against the highest international standards.

Provide the highest quality learning and teaching environment for the greater wellbeing of our students.

Produce graduates fully equipped to achieve the highest personal and professional standards.

Make a significant, sustainable and socially responsible contribution to Scotland, the UK and the world, promoting health and economic and cultural wellbeing.

As a great civic university, Edinburgh especially values its intellectual and economic relationship with the Scottish community that forms its base and provides the foundation from which it will continue to look to the widest international horizons, enriching both itself and Scotland.

Рубрика: Курсы/Образование/Постдоки | Добавить комментарий

CFP: ECIR 2018

В этом году я не успела ничего подать, а мои позапрошлогодние впечатление от ECIR можно найти вот тут. Надо будет в этом году постараться поучаствовать.

ECIR 2018 — European Conference on Information RetrievalCall for
long/short papers & demo

Grenoble, France — March 26-29, 2018



Full Papers, Short Papers and Demonstrations

We are seeking the submission of high-quality and original full papers,
short papers and demos. Submissions will be reviewed by experts on the
basis of the originality of the work, the validity of the results,
chosen methodology, writing quality and the overall contribution to the
field of IR. Short Paper submissions addressing any of the areas
identified in the conference topics are also invited. Authors are
encouraged to describe work in progress and late-breaking research
results. Demonstrations present research prototypes or operational
systems. They provide opportunities to exchange ideas gained from
implementing IR systems and to obtain feedback from expert users.
Demonstration submissions are welcome in any of the conference topic areas.

Reproducible IR Research Track

We are happy to announce that the Reproducible IR Research Track
introduced at ECIR 2015 will continue for ECIR 2018. Reproducibility is
critical for establishing reliable, referenceable and extensible
research for the future. Experimental papers are therefore most useful
when their results can be tested and generalised by peers. This track
specifically invites submission of papers reproducing a single or a
group of papers, from a third-party where you have *NOT* been directly
involved (e.g., *not* been an author or a collaborator). Emphasise your
motivation for selecting the paper/papers, the process of how results
have been attempted to be reproduced (successful or not), the
communication that was necessary to gather all information, the
potential difficulties encountered and the result of the process. A
successful reproduction of the work is not a requirement, but it is
important to provide a clear and rigid evaluation of the process to
allow lessons to be learned for the future.

Читать далее

Рубрика: Конференции | Добавить комментарий

магистерская программа Теоретическая лингвистика и описание языка (ВШЭ)

Магистерская программа Теоретическая лингвистика и описание языка продолжает набор магистрантов. Обучение ведется на английском языке, форма обучения: очная, вечерняя. Количество бюджетных мест — 20.
Отличительными чертами программы является включение междисциплинарного компонента (курсы экспериментальной и антропологической лингвистики, социолингвистики и инструментальной фонетики), акцент на анализе языковых данных (курс методов и практики статистического анализа), миникурсы по грамматике разноструктурных языков, а также возможность активно сочетать учебные курсы с самостоятельной исследовательской деятельностью. В рамках программы проводятся лекции и курсы приглашенных специалистов; осенью 2017 г. запланированы курсы Джоханны Николс (Johanna Nichols, University of California, Berkeley) и Стефана Гриса (Stefan Gries, University of California, Santa Barbara). Институциональными научными партнерами программы являются, в рамках НИУ ВШЭ, Лаборатория нейролингвистики, Лаборатория языковой конвергенции и Лаборатория языков Кавказа. У программы установлены образовательные и научно-исследовательские контакты с Полярным университетом Тромсе и другими европейскими университетами.
Школа лингвистики НИУ ВШЭ, на базе которой существует программа — центр современных лингвистических исследований самой широкой тематики, от корпусных исследований исторической грамматики и лексики русского языка до акустического анализа консонантных систем языков Дагестана. Магистры имеют возможность активно участвовать в исследовательских проектах Школы и ее партнеров, как теоретического, так и более прикладного характера. Один из ближайших учебных проектов Школы лингвистики — Летняя школа по лексике и типологии.
Подробнее о магистерской программе можно прочитать здесь. Обратите внимание — в рамках приемной кампании 2017 г. срок подачи документов заканчивается 20 июля.
Рубрика: Курсы/Образование/Постдоки | Добавить комментарий

Linked Data for Information Extraction LD4IE @ISWC2017

LD4IE 2017
The 5th international Workshop on Linked Data for Information Extraction
in conjunction with the 16th International Semantic Web Conference (ISWC 2017)
Vienna, Austria, October 21-25, 2017

Workshop website:
Twitter: @LD4IE #LD4IE #LD4IE2017
Facebook page: Ld4ie2017 (at

*************** Important Dates ***************

Abstract submission deadline: July 14, 2017
Paper submission deadline: July 21, 2017
Acceptance Notification: August 24, 2017
Camera-ready versions: September 1, 2017
Workshop date: to be announced (October 21-22, 2017)

*************** Call for Papers ***************

This workshop focuses on the exploitation of Linked Data for Web Scale Information Extraction (IE), which concerns extracting structured knowledge from unstructured/semi-structured documents on the Web.
One of the major bottlenecks for the current state of the art in IE is the availability of learning materials (e.g., seed data, training corpora), which, typically are manually created and are expensive to build and maintain.
Linked Data (LD) defines best practices for exposing, sharing, and connecting data, information, and knowledge on the Semantic Web using uniform means such as URIs and RDF.
It has so far been created a gigantic knowledge source of Linked Open Data (LOD), which constitutes a mine of learning materials for IE.
However, the massive quantity requires efficient learning algorithms and the unguaranteed quality of data requires robust methods to handle redundancy and noise.
LD4IE intends to gather researchers and practitioners to address multiple challenges arising from the usage of LD as learning material for IE tasks, focusing on (i) modelling user defined extraction tasks using LD; (ii) gathering learning materials from LD assuring quality (training data selection, cleaning, feature selection etc.); (ii) robust learning algorithms for handling LD; (iv) publishing IE results to the LOD cloud.

*************** Research Topics ***************

Topics of interest include, but are not limited to:

* Modelling Extraction Tasks
** extracting knowledge patterns for task modelling
** user friendly approaches for querying linked data

* Information Extraction
** selecting relevant portions of LOD as training data
** selecting relevant knowledge resources from linked data
** IE methods robust to noise in training data
** Information Extractions tasks/applications exploiting LOD (Wrapper induction, Table interpretation, IE from unstructured data, Named Entity Recognition, …)
** Domain specific IE consuming and producing LOD (social data, scholarly data, health data, …)
** publishing information extraction results as Linked Data
** linking extracted information to existing LOD datasets

* Linked Data for Learning
** assessing the quality of LOD data for training
** select optimal subset of LOD to seed learning
** managing incompleteness, noise, and uncertainty of LOD
** scalable learning methods
** pattern extraction from LOD

*************** Submission ********************

All submissions must be written in English.
We accept the following formats of submissions:
Full paper with a maximum of 12 pages including references.
Short paper with a maximum of 6 pages including references.

Two formats are possible for the submission: PDF and HTML.

PDF submissions must be formatted according to the information for LNCS Authors (

We would like to encourage you to submit your paper as HTML, in which case you need to submit a zip archive containing an HTML file and all used resources.
If you are new to HTML submission these are good places to start:
* dokieli ( is a clientside editor for decentralised article publishing, annotations and social interactions. It is compliant with the Linked Research ( initiative. Example papers using LNCS and ACM: and on website
* Research Articles in Simplified HTML (RASH) format: documentation and stylesheets at

In order to check if your HTML submission is compliant with the page limit constraint, please use one of the LNCS layouts and printing/storing it as PDF.

Please submit your contributions electronically in PDF or HTML format to EasyChair at
When submitting your paper, select the appropriate topic between:
* Research — long paper
* Research — short paper

Accepted papers will be published online via CEUR-WS.

Рубрика: Конференции | Добавить комментарий

Johns Hopkins University: Postdoc Positions

The Center for Language and Speech Processing (CLSP) at the Johns
Hopkins University seeks applicants for postdoctoral fellowship
positions in speech and language processing, including the areas of
natural language processing, machine learning and health
informatics. Applicants must have a Ph.D. in a relevant discipline and
a strong research record.

Possible research topics include:
— Trend Detection in Social Media
— Analysis of Clinical Medical Text
— Topic Identification and Analysis from Text
— Broadly Multilingual Learning of Morphology and Low-Resource Machine
— NLP and Machine Learning for Clinical Data Analysis

Host faculty include:
Mark Dredze, David Yarowsky, Jason Eisner, Sanjeev Khudanpur, Benjamin
Van Durme

Details and application information:

Рубрика: Вакансии/Стажировки, Курсы/Образование/Постдоки | Добавить комментарий