Max-Planck-Institut für Informatik
max planck institut
informatik
mpii logo Minerva of the Max Planck Society
 

MPI-INF or MPI-SWS or Local Campus Event Calendar

<< Previous Entry Next Entry >> New Event Entry Edit this Entry Login to DB (to update, delete)
What and Who
Title:Approaches for Word Representations
Speaker:Vinh Thinh Ho
coming from:International Max Planck Research School for Computer Science - IMPRS
Speakers Bio:IMPRS-CS Master student
Event Type:IMPRS Research Seminar
Visibility:D1, D2, D3, D4, D5, SWS, RG1, MMCI
We use this to send out email in the morning.
Level:Public Audience
Language:English
Date, Time and Location
Date:Monday, 23 October 2017
Time:12:00
Duration:45 Minutes
Location:Saarbrücken
Building:E1 4
Room:022
Abstract
Understanding the meaning of word is considered an important task in natural language processing. It could be then applied in a large number of
fields such as speech recognition, information retrieval, machine translate, or question answering. In this talk, we will go through some
approaches of word representation, which includes:


• Learn distributed representation of words. The main idea is to represent word as a vector of features that has the ability to group similar words,
distinguish different words and even more further tasks. These vectors are typically learned using neural network. We will go through some common
neural network models used to learn these feature vectors.

• We will also focus on the usage of Skip-gram model in learning distributed representation of words, how the model works and some methods
which are used to improve the performance of this model. We then discuss two extensions to speed up the learning phase of Skip-gram model by
sub-sampling frequent words, and help the model to learn phrase vectors by merging words based on a simple data-driven approach.

• Discuss briefly a frequency based approach to represent words, which is called distributional representation. This representation is efficient for
measuring the similarity between individual words.

Contact
Name(s):IMPRS-CS office
Phone:0681 93251800
EMail:--email address not disclosed on the web
Video Broadcast
Video Broadcast:NoTo Location:
Tags, Category, Keywords and additional notes
Note:
Attachments, File(s):

Created by:Stephanie Jörg/MPI-INF, 10/20/2017 11:57 AMLast modified by:Uwe Brahm/MPII/DE, 10/23/2017 07:00 AM
  • Stephanie Jörg, 10/20/2017 12:00 PM -- Created document.