elixir-reading-group team mailing list archive

[rathee <rathee@xxxxxx>]

-------- Original Message --------
Subject: Invited talk by Euna Jung: Semi-Siamese Bi-encoder Neural 
Ranking Model Using Lightweight Fine-Tuning
Date: 2022-05-17 09:28
From: Jurek Leonhardt <leonhardt@xxxxxx>
To: iai-announce@xxxxxx
Cc: Avishek Anand <avishek.anand@xxxxxxxxxx>, 정은아 <xlpczv@xxxxxxxxx>



Dear all,

this week we'll have an invited talk in our reading group. Euna Jung
from Seoul National University will present her recent WWW paper. This
talk should be especially interesting to everyone who works on neural
ranking or representation learning.

The talk takes place THIS THURSDAY (MAY 19TH) AT 8:30 CEST on Teams.

TITLE: Semi-Siamese Bi-encoder Neural Ranking Model Using Lightweight
Fine-Tuning

ABSTRACT:
For information retrieval, bi-encoder models that allow independent
query and document processing are practically beneficial. Traditional
BERT-based bi-encoders use Siamese models for both query and document,
where different characteristics of query and document might not be
effectively processed. We suggest a semi-Siamese structure for
bi-encoders using light-weight fine-tuning, improving the performance of
BERT-based bi-encoder models.

Meeting link:
https://teams.microsoft.com/l/meetup-join/19%3a7fe7541876ab4d7d9913a6cc16fc3c2f%40thread.tacv2/1592391869127?context=%7b%22Tid%22%3a%22c8875ac0-8ea4-46d5-abbc-720fe798fa01%22%2c%22Oid%22%3a%2204f4abca-6cd9-4b38-b249-c15122f25e24%22%7d


Best,

Jurek

--
Jurek Leonhardt

L3S Research Center / Forschungszentrum L3S
Leibniz Universität Hannover
Appelstraße 9a
30167 Hannover
Germany

https://www.l3s.de/~leonhardt/

Attachment: smime.p7s
Description: S/MIME cryptographic signature