QCDS2026

Welcome

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, a workshop co-located with the ECML/PKDD 2026 conference, 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.

QCDS 2026 is supported by project “Future Artificial Intelligence Research” (FAIR) and project “Strengthening the Italian RI for Social Mining and Big Data Analytics” (SoBigData.it), both funded by the European Union under the NextGenerationEU funding scheme (CUP B53D22000980006 and CUP B53C22001760006, respectively) and by the Agency for Science, Business Competitiveness, and Innovation of the Principality of Asturias in Spain (SEKUENS) through the project GRU-GIC-24-018.

Call for papers for the QCDS 2026 Workshop

We seek papers on any of the following topics (where QCDS stands for "Quantification AND/OR Classification under Dataset Shift"), 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 EasyChair.

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.

Important information

By submitting a paper the authors acknowledge that, in case of acceptance, one of them will register (according to the rules set by the ECML/PKDD 2026 organizers) and present the paper in presence at the workshop.

About the proceedings

The proceedings of the workshop will not be formally published, so as to allow authors to resubmit their work to other venues and so as to avoid putting the papers behind 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 (extended to June 12, 2026)
A/R notification deadline: June 29, 2026
Final copy submission deadline: July 10, 2026
Workshop: September 7, 2026

Organizers

Mirko Bunse

Lamarr Institute for Machine Learning and Artificial Intelligence, TU Dortmund University, Germany

Pablo González

Artificial Intelligence Center, University of Oviedo, Spain

Eyke Hüllermeier

Institute of Informatics at LMU Munich, Germany

Alejandro Moreo

Istituto di Scienza e Tecnologie dell’Informazione, Consiglio Nazionale delle Ricerche, Pisa, Italy

Fabrizio Sebastiani

Istituto di Scienza e Tecnologie dell’Informazione, Consiglio Nazionale delle Ricerche, Pisa, Italy

Program Committee

Gustavo Batista, University of New South Wales, AU
Clemens Damke, Ludwig Maximilian University of Munich, DE
Juan José del Coz, University of Oviedo, ES
Zahra Donyavi, University of New South Wales, AU
Andrea Esuli, Consiglio Nazionale delle Ricerche, IT
Cèsar Ferri, Universitat Politècnica de València, ES
Peter Flach, University of Bristol, UK
Devin Guillory, UC Berkeley, US
Barbara Hammer, University of Bielefeld, DE
Rafael Izbicki, Federal University of São Carlos, BR
Mira Jürgens, Gent University, BE
André Maletzke, Universidade Estadual do Oeste do Paraná, BR
Olaya Pérez-Mon, University of Oviedo, ES
Teodora Popordanovska, KU Leuven, BE
Marco Saerens, Catholic University of Louvain, BE
Tobias Schumacher, University of Mannheim, DE
Dirk Tasche, North-West University, SA
Lorenzo Volpi, Consiglio Nazionale delle Ricerche, IT
Willem Waegeman, Ghent University, BE

Program

When: Monday, September 7, 2026 (morning session).

This preliminary program might be subject to changes.

08:00	09:00	Conference Registration
09:00	09:30	Workshop Chairs' Introduction + mini-tutorial on ML under DS
		Session Chair: Pablo González
09:30	09:45	Zero-Shot Test-Time Canonicalization using Out-of-Distribution Scoring by Dominik Lindner, Johann Schmidt, Tom Siegl, Martin Becker and Sebastian Stober
09:45	10:00	A Non-Aggregative Quantification Method for Graph Nodes by Alessio Micheli, Alejandro Moreo, Marco Podda, Fabrizio Sebastiani and Domenico Tortorella
10:00	10:15	The Fragility of Value in Customer Churn: Predict-then-Optimize vs. Predict-and-Optimize Under Dataset Shift by Shimanto Rahman, Bram Janssens and Matthias Bogaert
10:15	10:30	Dataset Shifts in Temporal Quantification by João Pedro Ortega, Luiz Fernando Luth Junior, Daniel Zonta Ojeda, André Maletzke and Mirko Bunse
10:30	11:00	Coffee Break
11:00	12:00	Poster Session (See list of posters below)
		Session Chair: Mirko Bunse
12:00	12:15	Uncertainty-Aware Classifier Accuracy Prediction under Prior Probability Shift by Lorenzo Volpi, Alejandro Moreo and Fabrizio Sebastiani
12:15	12:30	Uncertainty-aware Finite-sample Quantification by Clemens Damke and Eyke Hüllermeier
12:30	12:45	Decision Tree Leaf Binning with Expectation Maximization for Quantification by David Kaftan, George Corliss, Anna Bershteyn and Richard Povinelli
12:45	13:00	Out-of-Known-Safes Detection as a Benchmark for Runtime Monitors by Robbe De Groeve, Laurens Devos and Mathias Verbeke
13:00	14:00	Lunch

Posters

Gram-MMD: A Texture-Aware Metric for Image Realism Assessment
by Joé Napolitano and Pascal Nguyen
Assessing the Impact of Dataset Shift for Automatic Nephrops norvegicus Burrows Detection and Classification Through Deep Learning
by Oscar Papini, Marco Reggiannini, Enrico Cecapolli, Filippo Domenichetti, Lorenzo Zacchetti, Michela Martinelli and Gabriele Pieri
Pre-Generation LLM Failure Prediction Under Distribution Shift
by Yael Moros-Daval, Álvaro David Gómez Antón, Kexin Jiang-Chen, Cèsar Ferri, José Hernández-Orallo and Fernando Martínez-Plumed
On Binary Quantification Calibration
by Isabella Caroline Sachini Lorena, André Gustavo Maletzke and Willian Zalewski
Drift Happens: An Empirical Study of Neural Architecture Robustness to Temporal Distribution Shift
by Robin Holzinger and Riccardo Colletti
Fishing Gear Classification from GPS Trajectories: Characterising Geographic Dataset Shift across Four Regions
by Hamza Altarturi, Alex Tilley, Lorenzo Longobardi, Giacomo Gardella and Simone Franceschini
Network Quantification on Neuromorphic Hardware: Proof-of-concept with Randomized Ising Models
by Maria Grazia Berni, Alessio Micheli, Marco Podda and Domenico Tortorella
Beyond Single-Objective Accuracy: A Longitudinal Evaluation of Tabular Foundation Models under Dataset Shift
by João Marcos Campos, Rafael Gomes, Diogo Chaves, Hugo Rocha, Mariangela Cherchiglia, Leonardo Rocha, Wagner Meira, and Marcos Gonçalves

Special Issue of the Data Mining and Knowledge Discovery journal on QCDS