Fwd: Séminaire AMA

[Radu Horaud]

Two interesting seminars… Don’t ask me for the location.

Begin forwarded message:

From: VINCENT BRAULT <vincent...@univ-grenoble-alpes.fr>

Subject: Fwd: Séminaire AMA

Date: March 7, 2018 at 3:14:11 PM GMT+1

To: ljk-pro...@univ-grenoble-alpes.fr, angelika studeny <angelika...@gmail.com>, Stephanie Berger <Stephani...@univ-grenoble-alpes.fr>, clementine prieur <clementi...@imag.fr>, rene blacher <rene.b...@imag.fr>, Herve Guiol <Herve...@imag.fr>, Olivier Francois <Olivier....@imag.fr>, Cecile Amblard <cecile....@imag.fr>, Michael Blum <michae...@imag.fr>, desbat <Laurent...@imag.fr>, picard <pic...@biomserv.univ-lyon1.fr>, olivier renault <olivier...@iut2.upmf-grenoble.fr>, Hidde De Jong <Hidde....@inrialpes.fr>, eugenio cinquemani <eugenio.c...@inria.fr>, nicolas le-bihan <nicolas....@gipsa-lab.grenoble-inp.fr>, Sophie Achard <sophie...@gipsa-lab.grenoble-inp.fr>, eric gaussier <Eric.G...@imag.fr>, pierre-olivier amblard <pierre-oliv...@gipsa-lab.grenoble-inp.fr>, olivier michel <olivier...@gipsa-lab.grenoble-inp.fr>, radu horaud <radu....@inrialpes.fr>, Didier Piau <Didie...@ujf-grenoble.fr>, Leuridan <Christoph...@ujf-grenoble.fr>, Raphael Rossignol <raphael....@ujf-grenoble.fr>, agnes coquio <agnes....@ujf-grenoble.fr>, benoit zupancic <benoit....@gmail.com>, mathilde grandjacques <mathilde.g...@g2elab.grenoble-inp.fr>, Juliette Blanchet <juliette...@ujf-grenoble.fr>, nicolas eckert <nicolas...@irstea.fr>, Laurent Condat <Laurent...@gipsa-lab.grenoble-inp.fr>, chichi leon <chichi...@hotmail.com>, Loren Coquille <lorenc...@gmail.com>, elamrireda pro <elamrir...@gmail.com>, Sylvain Lecorff <Sylvain...@math.u-psud.fr>, Irène Gannaz <irene....@insa-lyon.fr>, Vincent Beffara <Vincent...@ujf-grenoble.fr>, MYRIAM TAMI <myria...@univ-grenoble-alpes.fr>, CHARLOTTE LACLAU <charlott...@univ-grenoble-alpes.fr>, charline smadi-lasserre <charline.sm...@wanadoo.fr>, maria-belen heredia <maria-bel...@irstea.fr>, nicolas...@orange.fr, Massih-R...@univ-grenoble-alpes.fr, Glenn Cougoulat <glenn.c...@univ-grenoble-alpes.fr>, Simon Barthelmé <simon.b...@gipsa-lab.grenoble-inp.fr>

Bonjour,

voici l'annonce d'un séminaire AMA qui pourrait intéresser certains. Pour demain, n'oubliez pas que le séminaire commun aura lieu au rez-de-chaussée de la tour IRMA.

Bonne journée,

Caroline et Vincent

##############

# Message transféré

##############

Thursday the 15th: 13:00, Phong Nguyen and Ioannis Partalas from Expedia, room 1 (groundfloor).

The Expedia group is a leading global online travel agency, operating in more than 75 countries with more than 675M visits per month. One of our global tasks is to deliver the best shopping experience to travelers by connecting them with the best hotel suppliers and offers. In this talk, I will present the Expedia hotel sort algorithm and discuss its two main (conflicting) objectives; to maximize the customer conversion rate, with ideally personalization, while keeping at the same time good hotel margin or compensation. I will further discuss new deep learning approaches in recommender systems including auto-encoders, LSTM and GAN networks.

Phong Nguyen: I am a senior data scientist in Expedia where I am building the sort algorithm for various Expedia brands. Prior to that, I was a research assistant in the AI group of the University of Geneva where I did my PhD in meta-learning, trying to solve the task of data mining workflow selection and planning using meta-learning techniques.         

Every day, travelers query our search engine through free text searches to find relevant hotels or other products that would fit their needs. In order to provide relevant information to the customer we should be able to correctly analyze and understand the query. The first step to fulfill this goal is to extract the relevant entities such as an amenity, a location or a hotel name. This task is also known in the domain of natural language processing as Named Entity Recognition (NER) and the goal is to identify the named entities that appear in textual segments and classify them according to a predefined set of categories.  In this talk I will present how the NER can be framed as a Learning to Search task which is a framework for solving structured prediction tasks. I will further present empirical results on an in-house dataset and comparisons with state-of-the-art methods.

Ioannis Partalas works as machine learning scientist in Expedia group in various projects in content and revenue management. He was previously working as data scientist in the Viseo group (France) on NLP projects and prior to that he was a post-doctoral researcher in AMA team at University of Grenoble working on large-scale classification.

--

---------------------------------------------
L'urgent est fait
L'impossible est en cours
Pour les miracles, prévoir un délai
---------------------------------------------

-----------------------
Radu HORAUD  

https://team.inria.fr/perception/team-members/radu-patrice-horaud/
INRIA Grenoble Rhone-Alpes 
office (H218): +33 (0)4 76 61 52 26 

Assistant: Nathalie Gillot (Nathali...@inria.fr)