of the tutorial that we give as a part of the workshop program. Workshop Proceedings htp:/ceur-ws.org CEUR Workshop Proceedings (CEUR-WS.org) ISN1613-073
Methods of machine learning approach their limits whenever training data of a high quality are scarce. The potential reasons for data scarcity are manifold: limited capabilities of human supervisors and processing systems, a need for early predictions which can later be refined, or transfer settings where the only available data stem from a diferent learning task. Situations like these demand methods that improve the overall life-cycle of machine learning models, including interactions with human supervisors, interactions with other processing systems, and adaptations to diferent forms of data that become available at diferent points in time. This demand includes techniques for evaluating the impact of additional resources (e.g., data) on the learning process; strategies for actively selecting information to be processed or queried; techniques for reusing knowledge over time, across diferent domains or tasks, by recognizing similarities and by adapting to changes; and methods for efectively using diferent types of information, like labeled and unlabeled data, constraints, and knowledge. Techniques of this kind are being investigated, for example, in the areas of adaptive, active, semi-supervised, and transfer learning. While these investigations often happen in isolation of each other, real use nEvelop-O CEUR
i cases of machine learning require interactive and adaptive systems that operate under changing conditions, addressing the challenges of volume, velocity, and variability of the data.
This workshop stimulates research on systems that combine multiple areas of interactive and adaptive machine learning, by bringing together researchers and practitioners from these diferent areas. We welcome contributions that present a novel problem, propose a new approach, report practical experience with such a system, or raise open questions for the research community. This edition of the Interactive Adaptive Learning workshop, which is co-located with ECML-PKDD, continues a successful series of events, including a workshop & tutorial at ECML-PKDD 2017 in Skopje, a tutorial at IJCNN 2018 in Rio, a workshop at ECML-PKDD 2018 in Dublin, a workshop & tutorial at ECML-PKDD 2019 in Würzburg, a workshop at the virtual ECML-PKDD 2020, a workshop at the virtual ECML-PKDD 2021, and a workshop at ECML-PKDD 2022 in Grenoble.
This year, we accepted 8 papers out of 11 submissions for their publication in these workshop proceedings. The authors discuss topics such as deep active learning, meta-learning, fairness, active feature acquisition, and multiple applications of interactive adaptive learning. In addition to these contributions, we publish an extended abstract of a tutorial that belongs to the workshop program. We thank all authors for their valuable submissions and all members of the program committee for their great support.
Mirko Bunse Barbara Hammer Georg Krempl Vincent Lemaire Alaa Tharwat Amal Saadallah
Bielefeld University, Germany Utrecht University, Netherlands Orange Innovation, France Fachhochschule Bielefeld, Germany TU Dortmund University, Germany
Andreas Holzinger Daniel Kottke Robi Polikar Bernhard Sick
University of Natural Resources and Life Sciences Vienna, Austria Deutsche Bahn, Germany Rowan University, USA University of Kassel, Germany