LIDTA 2022 @ ECML/PKDD 2022

The problem of imbalanced domain learning has been thoroughly studied in the last two decades, with a specific focus on classification tasks. However, the research community has started to address this problem in other contexts such as regression, ordinal classification, multi-label and multi-class classification, association rules mining, multi-instance learning, data streams, time-series and spatio-temporal forecasting, text mining and multimodal data. Clearly, the research community recognises that imbalanced domains are a broad and important problem. Such a context poses important challenges for both supervised and unsupervised learning tasks, in an increasing number of real-world applications.

Tackling the issues raised by imbalanced domains is crucial to both academia and industry. To researchers, it is an opportunity to develop more adaptable and robust systems/approaches for very complex tasks. These tasks are, in many cases, those that industry is already facing today. These are very diverse and include the ability to prevent fraud, to anticipate catastrophes, and in general to enable a more preemptive action in an increasingly fast-paced world.

This workshop proposal focuses on providing a significant contribution to the problems of learning with imbalanced domains, aiming to increase the interest and the contributions to solving its challenges. The workshop invites inter-disciplinary contributions to tackle the problems that many real-world domains face nowadays. With the growing attention that this challenge has collected, it is crucial to promote its further development in order to tackle its theoretical and application challenges.

The research topics of interest to LIDTA'2022 workshop include (but are not limited to) the following:

• Foundations of learning in imbalanced domains


• Probabilistic and statistical models


• New knowledge discovery theories and models


• Probabilistic and statistical models


• New knowledge discovery theories and models


• Deep learning


• Handling imbalanced big data


• One-class learning


• Learning with non i.i.d. data


• Rare event detection in classification tasks


• New approaches for data pre-processing (e.g. resampling strategies)


• Post-processing approaches


• Sampling approaches


• Feature selection and feature transformation


• Evaluation metrics and methodologies


• Ensemble methods


• Instance hardness

• Knowledge discovery and machine learning in imbalanced domains


• Classification, ordinal classification


• Regression


• Data streams and time series forecasting


• Clustering


• Adaptive learning and algorithm-level approaches


• Multi-label, multi-instance, sequence and association rules mining


• Active learning


• Spatial and spatio-temporal learning


• Text and image mining


• Multi-modal learning


• Predictive Maintenance


• Automated machine learning


• Energy-efficiency

• Applications in imbalanced domains


• Health applications (e.g. medical imaging)


• Fraud detection (e.g. finance, credit and online banking)


• Anomaly detection (e.g. industry, intrusion detection, privacy and security)


• Environmental applications (e.g. meteorology, biology, oil spill detection)


• Social media applications (e.g. popularity prediction, recommender systems)


• Fake news detection and disinformation, deep fake classification


• Other real world applications and case studies

• Submission Deadline (extended): Friday, July 8, 2022


• Notification of Acceptance: Monday, July 18, 2022


• Camera-ready Deadline: Friday, August 5, 2022



• ECML/PKDD 2022: 19-23 September, 2022


• LIDTA 2022: 23 September, 2022

• 10h30: Paper Presentation I

- I. Antoniadis, V. Vercruyssen and J. Davis. Systematic Evaluation of CASH Search Strategies for Unsupervised Anomaly Detection
- S. De Block and J. Bekker. Bagging Propensity Weighting: A Robust method for biased PU Learning
- X. Yue Song, N. Dao and P. Branco. DistSMOGN: Distributed SMOGN for Imbalanced Regression Problems
- C. Ortega Vazquez, J. De Weerdt and S. Vanden Broucke. The Hidden Cost of Fraud: An Instance-Dependent Cost-Sensitive Approach for Positive and Unlabeled Learning
- Y. Shi, T. Valizadehaslani, J. Wang, P. Ren, Y. Zhang, M. Hu, L. Zhao and H. Liang. Improving Imbalanced Learning by Pre-finetuning with Data Augmentation
- J. da Silva Freitas Júnior and P. Henrique Pisani. Performance and model complexity on imbalanced datasets using resampling and cost-sensitive algorithms
- A. Wojciechowski and M. Lango. Adversarial oversampling for multi-class imbalanced data classification with convolutional neural networks
- T. Bonnier and B. Bosch. Assessing the Robustness of Ordinal Classifiers against Imbalanced and Shifting Distributions


• 12h30: Lightning Talks

- E. Rushe and B. Mac Namee. Deep Contextual Novelty Detection with Context Prediction
- B. Bauvin, D. Benielli, S. Koço, C. Capponi and J. Corbeil. Integrating and reporting full multi-view supervised learning experiments using SuMMIT
- S. Patricio Agostinho and J. Mendes Moreira. Probabilistic Metric to measure the imbalance in multi-class classifiers
- A. Mohammed Bouayed, S. Deslauriers-Gauthier, M. Zucchelli and R. Deriche. CNN and diffusion MRI's 4th degree rotational invariants for Alzheimer’s disease identification
- J. Komorniczak, P. Ksieniewicz and M. Wozniak. Data complexity and classification accuracy correlation in oversampling algorithms
- J. Stefanowski and A. Lipska. The Influence of Multiple Classes on Learning from Imbalanced Data Streams


• 13h00: Keynote Talk: Eirini Ntoutsi, Fairness under class-imbalance

Class imbalance is a common problem in machine learning which, if not tackled properly, might lead to poor performance in the minority class despite the overall good model performance. Another form of imbalance that has received much attention over the last years is that of population imbalance w.r.t. some protected attribute(s) like gender or/and race. Algorithms trained upon such data can affect different population subgroups differently, leading to bias and discrimination. The domain of fairness-aware machine learning focuses on methods and algorithms for understanding bias, mitigating bias, and accounting for bias in AI/ML models. However, most fairness-aware learning approaches ignore the fact that datasets in this domain are often imbalanced, leading to unfair decisions towards the minority class. In this talk, I will argue that fairness and class imbalance challenges should be tackled jointly, and I will present recent approaches in this direction.

• Colin Bellinger, University of Alberta


• Chris Drummond, NRC Institute for Information Technology


• Inês Dutra, DCC - Faculty of Sciences, University of Porto


• Alberto Fernández, Granada University


• Mikel Galar, Universidad Pública de Navarra


• Salvador Garcia, University of Granada


• Raji Ghawi, Technical University of Munich


• Bartosz Krawczyk, Virginia Commonwealth University


• Leandro Minku, University of Birmingham


• Rita Ribeiro, DCC - Faculty of Sciences, University of Porto


• Marina Sokolova, University of Ottawa


• Jerzy Stefanowski, Poznan University of Technology


• Gary Weiss, Fordham University

Proceedings
All accepted papers will be included in the workshop proceedings, published as a volume in Proceedings of Machine Learning Research (PMLR).

• - For each accepted paper, a presentation slot of 15 minutes is provided.
• - The maximum length for papers is 14 pages. Papers not respecting such limit will be rejected.


• - All submissions must be written in English and follow the PMLR format. Instructions for authors and style files may be found here, with pmlr-sample.


• - All submissions will be reviewed by the Program Committee using a double-blind method. As such, it is required that no personal information or reference to the authors should be introduced in the submitted paper.


• - Papers that have already been accepted or are currently under review for other workshops, conferences, or journals will not be considered.
• - Submissions will be evaluated concerning their technical quality, relevance, significance, originality and clarity.
• - At least one author of each accepted paper must attend the workshop and present the paper.





• Submit your paper!

• Nuno Moniz | INESC TEC / University of Porto, Portugal



• Paula Branco | University of Ottawa, Canada



• Luís Torgo | Dalhousie University, Canada



• Nathalie Japkowicz | American University, USA



• Michał Woźniak | Wroclaw University of Science and Technology, Poland



• Shuo Wang | University of Birmingham, UK