Unlabeled Set

Unlabeled datasets, containing data without associated class labels, are a crucial resource in machine learning, enabling the development of models with improved accuracy and robustness, especially when labeled data is scarce or expensive. Current research focuses on leveraging unlabeled data through techniques like self-training, where models iteratively predict labels and refine their performance, and ensemble methods that combine predictions from multiple models trained on different subsets of the unlabeled data. These approaches aim to address challenges such as calibration issues, semantic drift, and overfitting, ultimately leading to more reliable and efficient machine learning models across various applications, including text classification, image recognition, and even automated scoring of complex assessments.

Papers

December 31, 2023

Neural Networks Against (and For) Self-Training: Classification with Small Labeled and Large Unlabeled Sets
Payam Karisani
Language Model Neural Network Classification Code Pseudo Label Pseudo Labeling Noisy Pseudo Label Semi Supervised Text Classification Unlabeled Set

October 4, 2023

Something for (almost) nothing: Improving deep ensemble calibration using unlabeled data
Konstantinos Pitas, Julyan Arbel
Deep Learning Unlabeled Data Deep Ensemble Ensemble Diversity Small Training Unlabeled Set

September 19, 2022

A Closer Look at Novel Class Discovery from the Labeled Set
Ziyun Li, Jona Otholt, Ben Dai, Di hu, Christoph Meinel, Haojin Yang
Supervised ImageNet Glance Annotation Novel Class Discovery Unlabeled Dataset Unlabeled Set

July 4, 2022

Multi-class Classification from Multiple Unlabeled Datasets with Partial Risk Regularization
Yuting Tang, Nan Lu, Tianyi Zhang, Masashi Sugiyama
Multi Class Classification Classification Error Supervised Deep Learning Unlabeled Set

January 31, 2022

Uncertainty-aware Pseudo-label Selection for Positive-Unlabeled Learning
Emilio Dorigatti, Jann Goschenhofer, Benjamin Schubert, Mina Rezaei, Bernd Bischl
Binary Classifier Positive Unlabeled Learning Unlabeled Learning Imbalanced Dataset Unlabeled Set Uncertainty Aware Pseudo Label

January 10, 2022

Handwriting recognition and automatic scoring for descriptive answers in Japanese language tests
Hung Tuan Nguyen, Cuong Tuan Nguyen, Haruki Oka, Tsunenori Ishioka, Masaki Nakagawa
Human Understanding Handwriting Recognition Handwritten Text Automatic Scoring Handwriting Dataset Language Test Unlabeled Set

Unlabeled Set

Papers

Neural Networks Against (and For) Self-Training: Classification with Small Labeled and Large Unlabeled Sets

Something for (almost) nothing: Improving deep ensemble calibration using unlabeled data

A Closer Look at Novel Class Discovery from the Labeled Set

Multi-class Classification from Multiple Unlabeled Datasets with Partial Risk Regularization

Uncertainty-aware Pseudo-label Selection for Positive-Unlabeled Learning

Handwriting recognition and automatic scoring for descriptive answers in Japanese language tests