Call for Papers: 1st International Workshop on Quantification and Classification under Dataset Shift (QCDS 2026)
Pablo González
Quantification and classification are two supervised learning tasks that, in real applications, are often hampered by changes in data distributions, i.e., by dataset shift. The two tasks differ in what trained models predict: while a classifier predicts the class label of each individual data point, a quantifier predicts the prevalence (i.e., relative frequency) of each class in a set of unlabelled data points. Both classifiers and quantifiers can suffer if dataset shift is at play, at least as long as they are not designed to handle the current type of shift robustly. Research has shown that a quantifier robust to dataset shift can facilitate robust classification, and a classifier robust to dataset shift can facilitate robust quantification.
QCDS 2026 aims to engage the diverse expertise of the ECML/PKDD community; as dataset shift remains a fundamental challenge in real-world deployments, understanding the interplay between classification and quantification is more critical than ever. This workshop provides a collaborative forum for researchers and practitioners to bridge the gap between these two vital fields, to share breakthroughs in machine learning methods robust to dataset shift, and to explore emerging applications.
QCDS 2026 is a follow-up of the “Learning to Quantify” (LQ) workshop series; while the LQ workshops concentrated exclusively on quantification under dataset shift, QCDS 2026 has a broadened scope, and also encompasses classification under dataset shift and how, when dataset shift is at play, quantification and classification may bring mutual benefit.
We seek papers on any of the following topics, which will form the main themes of the QCDS 2026 workshop:
- Binary, multiclass, multilabel, and ordinal QCDS
- Semi-supervised / transductive QCDS
- Multi-instance learning / learning for set-structured data
- Deep learning for QCDS
- Characterizations of dataset shift, and their impact on QCDS
- Detecting and measuring (different types of) dataset shift
- Evaluation measures and protocols for QCDS
- Improving classification under dataset shift via quantification
- Classifier calibration under dataset shift
- New datasets and applications for evaluating QCDS
and other topics of relevance to QCDS. Two categories of papers are of interest:
- Papers reporting original, unpublished research;
- Papers {published in 2026 / currently under submission / accepted in 2026} at other {workshops / conferences / journals}, provided this double submission does not violate the rules of these {workshops / conferences / journals}.
Submission
Papers should be submitted (specifying which of the two above categories they belong to) via the EasyChair system at
Papers should be formatted according to Springer’s LNCS template, and should be up to 16 pages (including references) in length; however, this is just the upper bound, and contributions of any length up to this bound will be considered.
Other information
IMPORTANT: By submitting a paper the authors commit, in case of acceptance, to have one of them register (according to the rules set by the ECML/PKDD 2026 organizers) and present the paper at the workshop. The proceedings of the workshop will not be formally published, so as to allow authors to resubmit their work to other conferences and so as to avoid paywalls. Informal proceedings will be published on the workshop website; however, for each accepted paper, it will be left at the discretion of the authors to decide whether to contribute their paper or not to these proceedings.
Important dates (all deadlines are 23:59 AoE)
- Paper submission deadline: June 5, 2026
- A/R notification deadline: June 29, 2026
- Final copy submission deadline: July 10, 2026
- Workshop: September 7 (likely) or 11 (unlikely), 2026