[UAI] [news] The First Continual Semi-Supervised Learning Challenge @ IJCAI 2021

[Fabio Cuzzolin]

*The First Continual Semi-Supervised Learning Challenge*

*Call for participation*

The Challenge is organised as part of the upcoming IJCAI 2021 *First
International Workshop on Continual Semi-Supervised Learning*

https://sites.google.com/view/sscl-workshop-ijcai-2021/

*Aim of the Workshop*

Whereas continual learning has recently attracted much attention in the
machine learning community, the focus has been mainly on preventing the
model updated in the light of new data from ‘catastrophically forgetting’
its initial knowledge and abilities. This, however, is in stark contrast
with common real-world situations in which an initial model is trained
using limited data, only to be later deployed without any additional
supervision. In these scenarios the goal is for the model to be
incrementally updated using the new (unlabelled) data, in order to adapt to
a target domain continually shifting over time.

The aim of this workshop is to formalise this new *continual* *semi-supervised
learning* paradigm, and to introduce it to the machine learning community
in order to mobilise effort in this direction. We present the first two
benchmark datasets for this problem, derived from significant computer
vision scenarios, and propose the first *Continual Semi-Supervised Learning
Challenges* to the research community.

*Problem Statement*

In semi-supervised continual learning, an initial training batch of data
points annotated with ground truth (class labels for classification
problems, or vectors of target values for regression ones) is available and
can be used to train an initial model. Then, however, the model is
incrementally updated by exploiting the information provided by a time
series of *unlabelled* data points, each of which is generated by a data
generating process (modelled, as typically assumed, by a probability
distribution) which may vary with time, without any artificial subdivision
into ‘tasks’.

*Challenges*

We propose both a *continual activity recognition *(CAR) challenge and
a *continual
crowd counting* (CCC) challenge.

https://sites.google.com/view/sscl-workshop-ijcai-2021/challenges

In the former, the aim is to devise a learning mechanism for updating a
baseline action recognition method (working at frame level) based on a data
stream of video frames, of which only the initial fraction is labelled (a
classification problem).

In the latter, the learning mechanism is applied to a baseline crowd
counting method, also working on a frame-by-frame basis, and exploits a
data stream of video frames of which only an initial fraction come with
ground truth attached in the form of a density map (a regression problem).

*Benchmark Datasets*

As a benchmark for the continual activity recognition challenge we have
created a *Continual Activity Recognition (CAR) dataset*, derived from a
fraction of the MEVA (Multiview Extended Video with Activities) activity
detection dataset (https://mevadata.org/). We selected a suitable set of 8
activity classes from the original list of 37, and annotated each frame in
15 video sequences, each composed by 3 clips originally from MEVA, with a
single class label.

Our CAR benchmark is thus composed of 15 sequences, broken down into three
groups:

· Five 15-minute-long sequences formed by three original videos which are
contiguous.

· Five 15-minute-long sequences formed by three videos separated by a short
time interval (5-20 minutes).

· Five 15-minute-long sequences formed by three original videos separated
by a long interval of time (hours or even days).

Each of these three evaluation settings is designed to simulate a different
mix of continuous and discrete dynamics of the domain distribution.

The raw video sequences are directly accessible from the Challenge website.

Our CCC benchmark is composed of 3 sequences, taken from existing crowd
counting datasets:

· A single 2,000 frame sequence from the Mall dataset.

· A single 2,000-frame sequence from the UCSD dataset.

· A 750-frame sequence from the Fudan-ShanghaiTech (FDST) dataset, composed
of 5 clips portraying the same scene each 150 frames long.

*Ground truth*

The ground truth for the CAR challenges (in the form of one activity label
per frame) was created by us, after selecting a subset of 8 activity
classes and revising the original annotation for the 45 video clips we
selected for inclusion.

The ground truth for the CCC challenges (in the form of a density map for
each frame) was generated by us for all three datasets following the
annotation protocol described in

https://github.com/svishwa/crowdcount-mcnn

The ground truth for both challenges will be released on the Challenge web
site according to the following schedule:

·Training and validation fold release: May 5 2021

·Test fold release: June 30 2021

·Submission of results: July 15 2021

·Announcement of results: July 31 2021

·Challenge event @ workshop: August 21-23 2021

*Tasks*

For each challenge we p

[UAI] [jobs] PhD student position in Learning and Decision Making for Autonomous Driving

[Kyrki Ville]

PhD student position in Learning and Decision Making for Autonomous Driving

Future robots need to possess human-like capabilities in areas such as 
perception, decision making, and reasoning. Intelligent Robotics group 
(irobotics.aalto.fi) at Aalto University’s School of Electrical Engineering 
works actively to develop intelligent robotic systems with a particular 
emphasis on methods and systems that cope with imperfect knowledge and 
uncertain sensors. The research environment provides excellent opportunities 
for open-minded co-operation with highly motivated research staff as well as 
with top national and international partners. The group is a key player in a 
university-wide autonomous vehicle initiative, providing access to an advanced 
autonomous driving platform. The group is affiliated with Finnish Center for AI 
(fcai.fi) and Aalto Center for Autonomous Systems (acas.fi). The group is 
highly international and the working language is English.

We seek for excellent PhD student candidates to help us study learning, 
decision making and control in autonomous driving. Potential research topics 
include embedding prescribed rules in learned policies, handling anomalous 
situations and learning predictive multi-agent models.

Requirements

A candidate should have a M.Sc. or equivalent degree in a suitable area (e.g. 
robotics, computer science, automation, applied math, control engineering), or 
be a M.Sc. student in a final phase of their studies. The candidate should have 
strong analytical and writing skills, and experience or genuine interest in 
robotics. Expertise in machine learning is valued. Candidates are expected to 
also have good programming and mathematical skills as well as be fluent in 
spoken and written English.

The appointment is available immediately but the timing is negotiable. The 
position is fixed-term. The length of PhD studies in Finland is 4 years. The 
salary is determined according to the salary system of Aalto University.  
Employment includes occupational health care and social security benefits.

How to apply

To apply for the position, read the full instructions and apply at 
https://www.aalto.fi/en/open-positions/phd-student-position-in-learning-and-decision-making-for-autonomous-driving

The call is open until 4 June, 2021, but we will start reviewing and 
interviewing candidates immediately. All applicants will be notified on the 
decisions.

Additional information

For additional information, please contact Ville Kyrki by e-mail 
ville.ky...@aalto.fi. Additional information in recruitment process related 
questions, please contact HR Coordinator Jaana Hänninen, 
jaana.hanni...@aalto.fi.

About Aalto University, Helsinki, and Finland

Aalto University (aalto.fi) is located in Finland which is among the best 
countries in the world according to many quality of life indicators, including 
being the happiest country in the world (UN study 2018 and The World Happiness 
Report 2019) and the safest country in the world (World Economic Forum report 
2017). Aalto University is the foremost university in Finland in Engineering, 
Design and Business. Less than a 15 minutes metro ride away from center of 
Helsinki, capital of Finland, Aalto offers access to rich cultural and social 
life to help maintain healthy work-life balance.

Aalto University is a community of bold thinkers where science and art meet 
technology and business. We are committed to identifying and solving grand 
societal challenges and building an innovative future. Aalto has six schools 
with nearly 11 000 students and a staff of more than 4000, of which 400 are 
professors. Diversity is part of who we are, and we actively work to ensure our 
community’s diversity and inclusiveness in the future as well. This is why we 
warmly encourage qualified candidates from all backgrounds to join our 
community.

For more information about living in Finland: 
http://www.aalto.fi/en/about/careers/international_staff/ 

___
uai mailing list
uai@engr.orst.edu
https://it.engineering.oregonstate.edu/mailman/listinfo/uai

[UAI] [CFP] Call for Abstracts and Call for the e-Business 2021 Doctoral Consortium (Submission Deadline :: 12th of May)

[calendarsites]

Dear Colleagues,

ICE-B 2021 is welcoming abstract submissions and also contributions to the
e-Business 2021 Doctoral Consortium, until the 12th of May.

ABSTRACT TRACK:

In order to ensure a proper and fair review of the abstracts, we encourage
authors to submit abstracts with clear insights and background of the work,
and to include complementing material that can help the panel understand its
contribution and relevance to the conference. Complementing material could
include, for example, previously published papers or videos related to the
abstract, to enrich the abstract submission. 

Abstracts that bring an appropriate complementing material are eligible for
oral or poster presentation; otherwise, they are eligible for poster
presentation only. Oral presentations are given a 20' minutes time slot.

Abstracts will be advertised in the conference program and in the conference
book of abstracts (in the printed and digital version), but not published

DOCTORAL CONSORTIUM:

The ICE-B 2021 Doctoral Consortium aims to improve the research of PhD
students and broaden their perspectives by giving them the opportunity to
share and develop their research ideas in a supportive environment, get
feedback from senior members, improve their communication skills, exchange
ideas, and build relationships with other PhD students.

Accepted contributions will be presented in a dedicated conference session
where participants will discuss their research ideas and results, and
receive constructive feedback from an audience consisting of their peers as
well as more senior experts in the field. In addition to that, students will
be invited to participate at poster sessions so that they also receive
comments of other participants in the conference.

 

IMPORTANT DATES:

Submission deadline: May 12, 2021

Notification of acceptance:   May 25, 2021

Camera Ready and Registration: June 3, 2021

 

 

For more information, you may send an email to:
 ice-b.secretar...@insticc.org or
visit the conference website:  
http://www.ice-business.org/

 

___
uai mailing list
uai@engr.orst.edu
https://it.engineering.oregonstate.edu/mailman/listinfo/uai

[UAI] [CFP] Call for Abstracts and Call for the Wireless Networks and Mobile Systems Doctoral Consortium (Submission Deadline :: 12th of May)

[calendarsites]

Dear Colleagues,

WINSYS 2021 is welcoming abstract submissions and also contributions to the
Wireless Networks and Mobile Systems 2021 Doctoral Consortium, until the
12th of May.

ABSTRACT TRACK:

In order to ensure a proper and fair review of the abstracts, we encourage
authors to submit abstracts with clear insights and background of the work,
and to include complementing material that can help the panel understand its
contribution and relevance to the conference. Complementing material could
include, for example, previously published papers or videos related to the
abstract, to enrich the abstract submission. 

Abstracts that bring an appropriate complementing material are eligible for
oral or poster presentation; otherwise, they are eligible for poster
presentation only. Oral presentations are given a 20' minutes time slot.

Abstracts will be advertised in the conference program and in the conference
book of abstracts (in the printed and digital version), but not published.

DOCTORAL CONSORTIUM:

The Wireless Networks and Mobile Systems 2021 Doctoral Consortium aims to
improve the research of PhD students and broaden their perspectives by
giving them the opportunity to share and develop their research ideas in a
supportive environment, get feedback from senior members, improve their
communication skills, exchange ideas, and build relationships with other PhD
students.

Accepted contributions will be presented in a dedicated conference session
where participants will discuss their research ideas and results, and
receive constructive feedback from an audience consisting of their peers as
well as more senior experts in the field. In addition to that, students will
be invited to participate at poster sessions so that they also receive
comments of other participants in the conference.

 

IMPORTANT DATES:

Submission deadline: May 12, 2021

Notification of acceptance:   May 25, 2021

Camera Ready and Registration: June 3, 2021

 

 

For more information, you may send an email to:
winsys.secretar...@insticc.org   or
visit the conference website: http://www.winsys.org

 

___
uai mailing list
uai@engr.orst.edu
https://it.engineering.oregonstate.edu/mailman/listinfo/uai