THE FIRST INTERNATIONAL WORKSHOP ON AUTONOMOUS LEARNING IN COMPLEX DECISION SITUATIONS
23 June 2023, Room 3, Gold Coast Convention and Exhibition Centre, Queensland, Australia
The aim of the workshop is to create an integrated and holistic computational foundation for a new research direction – autonomous learning in complex decision situations. We define a decision situation as complex if the data available for use in machine learning efforts is massive and/or uncertain and/or dynamic. Autonomous learning will advance the capability of machines to learn from complex situations and minimise human involvement in the learning process (such as to autonomously determine a threshold, a sample set, a source domain, a concept drift, and a policy).
BACKGROUND
Recently, we have seen several new successful developments, such as massive stream learning algorithms, and incremental and online learning for streaming data. These developments have demonstrated how autonomous learning can be used in some complex decision situations to contribute to the implementation of machine learning capability. We have also witnessed some compelling evidence of successful investigations on using the autonomous learning methodology to support real-time prediction and decision making in practice.
AIMS
With these observations, it is instructive, vital, and timely to offer a unified view of the current trends and form a broad forum for fundamental and applied research as well as the practical development of autonomous learning in complex decision situations for improving machine learning and data-driven decision support systems.
DETAILED DESCRIPTIONS
The workshop is centred around three main research focuses regarding autonomous learning (but not limited):
​
General automated machine learning (AutoML) targets to automatically select, compose, process, measure, and parametrise machine learning models. It reduces the demand for human experts and experience. Along this direction, this workshop will include (but not limited to) the following topics:
Model selection, hyper-parameter optimization, and model search
Neural architecture search
Automatic feature transformation to match algorithm requirements
Bayesian optimization for AutoML
Evolutionary algorithms for AutoML
Autonomous transfer learning (ATL) in massive and uncertain domains, aiming to autonomously measure the relation between source domains and other domains where data is insufficient, and to optimise the transfer process from source domains to other domains. Along this direction, this workshop will include (but not limited to) the following topics:
Integral probability metrics, f-divergences
Density ratio estimation
Multi-source domain adaptation
Multi-target domain adaptation
Domain adaptation under weak supervision
Autonomous drift learning (ADL) in massive and dynamic data streams, aiming to autonomously detect, trace the causes of, and adapt massive-stream concept drift and correlation drift to support decisions, given unpredictable stream pattern changes. Along this direction, this workshop will include (but not limited to) the following topics:
Integral probability metrics, f-divergences
Multi-stream concept drift detection
Multi-stream concept drift adaptation
Autonomous drift cause-tracing
SCHEDULE
08:20 - 12:05, 23 June, 2023 (Australian Eastern Standard Time (AEST), UTC +10)
08:20 - 10:00
SESSION ONE
Session Chair: Prof Jie Lu
In this session, Prof Lu will give opening remarks for the workshop, and one keynote talk and two invited talks will be presented.
08:20Â - 08:25
OPENING REMARKS
Host: Prof Jie Lu
Prof Lu will briefly introduce autonomous learning in complex decision situations and her ARC Laureate Project.
08:25Â - 09:10
KEYNOTE I
Presenter: Prof Xue Li (The University of Queensland)
Title: High-Order Reasoning with Large Language Model in Life-Critical Decisions
09:10Â - 10:00
INVITED FEATURED PRESENTATIONS (25MINS EACH)
Presenter: Dr Javier Andreu-Perez (University of Essex)
Title: From Neuroscience to Human-Centred Autonomous Intelligent Systems
​
Presenter: Dr Paul Darwen (James Cook University)
Title: Direction of the Difference between Bayesian Model Averaging and the Best-Fit Model on Scarce-Data Low-Correlation Churn Prediction
10:00 - 10:30
COFFEE BREAK
Coffee Break, Gallery
10:30 - 12:05
SESSION TWO
Session Chair: Prof Xin Yao
In this session, one keynote talk and one invited talk will be presented. In the end, Dr Feng Liu will organize a panel discussion, and Prof Yao will conclude the whole workshop and give closing remarks.
10:30 - 11:15
KEYNOTE II
Presenter: Prof Marley Vellasco (Pontifical Catholic University of Rio de Janeiro)
Title:Â Neural Architecture Search based on quantum-inspired evolutionary algorithm
11:15Â - 11:40
INVITED TALK
Presenter: Dr Zhen Fang (University of Technology Sydney)
Title: Autonomous Out-of-distribution Detection: Theory and Algorithm
11:40Â - 12:00
PANEL DISCUSSION
Host: Dr Feng Liu
Panel Members: Prof Jie Lu, Prof Xin Yao, Prof Xue Li, and Prof Marley Vellasco
Topic: Challenges and Opportunities of Autonomous Learning
12:00 - 12:05
CLOSING REMARKS
Host: Prof Xin Yao
Prof Xin Yao will conclude the whole workshop and give closing remarks.
SUBMISSION DETAILS
Single-Blind Reviewing
The review process for AutoL2023 will be single-blind, i.e. The authors will not know the identity of the Reviewers.
​
Paper Submission
The format of submissions should refer to the IJCNN2023 formatting instructions for the Conference Track. Four-page submissions will be considered for the poster, while eight-page submissions will be considered for the oral presentation. We receive the workshop submissions and camera-ready versions by email workshop.ijcnn.autol@gmail.com with the subject line AutoL-IJCNN2023-{paper name}.
AutoL2023 is a non-archival venue and there will be no published proceedings. The accepted submissions will be presented on the workshop website. Besides, we also welcome submissions to AutoL2023 that are under review at other venues, if the concurrent submissions are permitted. At least one author from each accepted submission must register for the workshop.
​
Manuscript Style Information
Only papers in PDF format will be accepted.
Paper Size: A4 (210mm x 297mm).
Paper Length: Each paper should have 4 to MAXIMUM 8 pages, including figures, tables and references.
Paper Formatting: double column, single spaced, #10 point Times Roman font. Please make sure to use the official IEEE style files provided above.
Note: Violations of any of the above specifications may result in rejection of your paper.
IMPORTANT DATES
Paper submission: 16 April 2023 (11:59 PM AoE) STRICT DEADLINE
Notification of acceptance: 1 May 2023
Camera-ready paper submission: May 28, 2023
Workshop Day: 23 June, 2023
ORGANISERS
GET IN TOUCH
If you have questions about the submission/registration process, don’t hesitate to reach out.