v263: Proceedings of CLDD 2024

2026-06-10 · Source: Proceedings of Machine Learning Research · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Data Science & Analytics · Depth: Expert, quick

Summary

Volume 263 of the "Proceedings of The Workshop on Classifier Learning from Difficult Data," held in Santiago de Compostella, Spain, on October 19-20, 2024, presents research on various challenges in classifier learning. Edited by Pawel Zyblewski, Manuel Grana, Ksieniewicz Pawel, and Leandro Minku, the workshop features papers addressing issues such as contaminated multi-channel signals for bioprosthesis control and visible-infrared person re-identification. Contributions also explore methods for speeding up deep neural network training, evaluating imbalanced data classifiers using tools like the "F_{\beta}-plot", and learning few-shot contrastive representations with "Silhouette Distance Loss". Further research focuses on detecting noisy labels with early stopped models and investigating the role of alignment in continual learning, highlighting diverse approaches to robust classification. This collection provides insights into advanced techniques for handling complex and challenging datasets in machine learning.

Key takeaway

This volume addresses critical challenges in classifier learning from difficult data, including noisy labels, imbalanced distributions, and multi-modal inputs. It introduces novel techniques like dual ensemble classifiers for contaminated signals, streaming DNN training with base-values, and the $F_{\beta}$-plot for imbalanced data evaluation. These advancements offer practical solutions for AI/ML practitioners and researchers building robust models in real-world, imperfect data environments.

Topics

• Classifier Learning
• Difficult Data
• Deep Neural Networks
• Imbalanced Data
• Few-Shot Learning

Best for: Computer Vision Engineer, Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• v154: Proceedings of LIDTA 2021 — The Third International Workshop on Learning with Imbalanced Domains: Theory and Applications, held on September 17, 2021, at ECML-PKDD, focused on advanced machine learning techniques for datasets wi…
• v27 — Volume 27 of the Proceedings of the ICML Workshop on Unsupervised and Transfer Learning, held on July 2, 2011, in Bellevue, Washington, USA, compiles significant research in these machine learning domains.
• Deep Active Re-Labeling: Toward Noise-Resilient Annotation Efficiency — Deep Active Re-Labeling improves active learning by strategically re-annotating noisy data to enhance efficiency and data quality.
• Exploring Data Augmentation and Resampling Strategies for Transformer-Based Models to Address Class Imbalance in AI Scoring of Scientific Explanations in NGSS Classroom — Targeted data augmentation significantly improves transformer-based model performance in automated scoring of imbalanced educational text data.
• Practical estimation of the optimal classification error with soft labels and calibration — Isotonic calibration enables consistent Bayes error estimation even with corrupted soft labels, provided their order is preserved.
• Correcting Class Imbalance in Prior-Data Fitted Networks for Tabular Classification — PFNs' class imbalance issues are best addressed by simple thresholding or downsampling due to their unique calibration.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Proceedings of Machine Learning Research.