ACM SIGCHI General Interest Announcements (Mailing List)


Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Condense Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"ACM SIGCHI General Interest Announcements (Mailing List)" <[log in to unmask]>
"Dr. Graham Healy" <[log in to unmask]>
Sun, 25 Jun 2017 15:39:45 +0100
text/plain; charset="UTF-8"
"Dr. Graham Healy" <[log in to unmask]>
text/plain (100 lines)
*Apologies for any cross posting*
CALL FOR TASK PARTICIPATION: NAILS (Neurally Augmented Image Labelling
Strategies) at NTCIR-13

Registration deadline: June 30th (extended)
NTCIR-13 Conference: 5-8 Dec 2017 (NII, Tokyo, Japan)


The goal of the NAILS (Neurally Augmented Image Labelling Strategies) task at
NTCIR-13 is to explore strategies for generating image annotations that use
features extracted from the EEG (Electroencephalography) readings of users
performing high-speed image search tasks. This is an emerging technique
that can potentially achieve high information throughput rates, however,
finding machine learning and feature processing strategies that work
optimally across different image-search task types and users can be

NAILS aims to explore these challenges in the form of a collaborative
evaluation task where participating organisations are given access to a
dataset of EEG responses for users completing a variety of image search
tasks. Participating organisations will train machine-learning prediction
model(s) on the provided training set of EEG responses, and later benchmark
these model(s) on a test set (where the ground truth is kept by the
NAILS organisers).
Participating organisations will be able to submit test set predictions
multiple times in order to refine their approach before submitting their
paper for later presentation at NTCIR in NII, Tokyo, Japan on the 5-8
December 2017.

Other important details:

   - Although this is a collaborative evaluation where participating
   organisation's machine-learning strategies will be ranked in terms of
   balanced accuracy, it is expected that many good signal
   processing/machine-learning solutions that may perform suboptimally to
   others in terms of accuracy alone may offer other advantages in terms of
   speed, model complexity, neurophysiological interpretability and/or cross-
   task/user applicability. *Contributions exploring such aspects of the
   dataset and/or tasks are highly encouraged.*

   - As we expect researchers from different research backgrounds who have
   experience in machine learning who wish to work with the data, we provide
   preprocessed data so that those who may be less familiar with working with
   EEG (such as filtering/processing) may directly progress to applying
   machine-learning techniques e.g. preprocessed multi-channel time-series and
   wavelet-based feature vectors are available.

   - We provide working code examples (python) of complete pipelines (from
   training to prediction submission) that leverage existing common
   machine-learning strategies

   - This in an interdisciplinary research area involving areas such as IR
   (Information Retrieval), BCI (Brain-computer Interfaces), ML (machine
   learning), HCI (Human-computer Interaction) and signal processing.

For more information (and to find related publications of the dataset)
please visit:

To get access to the data/participate in the evaluation, please complete
see the registration and user agreement forums:

Important Dates:
30 June 2017 - Task Registration Due
1 June to 15 Sept 2017 - Formal Run
20 Sep 2017 - Evaluation Results Release
1 Oct 2017 - Paper Due
1 Nov 2017 - Camera ready due
5-8 Dec 2017 - NTCIR-13 in NII, Tokyo

Kind Regards,
Dr. Graham Healy,
The Insight Centre for Data Analytics,
School of Computing,
Dublin City University,

Tel: 00 353 700 6840
e-mail: [log in to unmask] / [log in to unmask]

    For news of CHI books, courses & software, join CHI-RESOURCES
     mailto: [log in to unmask]

    To unsubscribe from CHI-ANNOUNCEMENTS send an email to
     mailto:[log in to unmask]

    For further details of CHI lists see