Learning From Noisy Large-Scale Datasets With Minimal Supervision

2017/01/06 by Andreas Veit, Neil Alldrin, Veit, Andreas +9 · 10 citations
Computer Science · Medicine · #Advanced Neural Network Applications #COVID-19 diagnosis using AI #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #cs.CV

paper · pdf · doi:10.48550/arxiv.1701.01619

CVPR 2017

openalex publication_date 2017/01/06 · arxiv created 2017/04/10 · arxiv updated 2017/04/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We present an approach to effectively use millions of images with noisy annotations in conjunction with a small subset of cleanly-annotated images to learn powerful image representations. One common approach to combine clean and noisy data is to first pre-train a network using the large noisy dataset and then fine-tune with the clean dataset. We show this approach does not fully leverage the information contained in the clean set. Thus, we demonstrate how to use the clean annotations to reduce the noise in the large dataset before fine-tuning the network using both the clean set and the full set with reduced noise. The approach comprises a multi-task network that jointly learns to clean noisy annotations and to accurately classify images. We evaluate our approach on the recently released Open Images dataset, containing ~9 million images, multiple annotations per image and over 6000 unique classes. For the small clean set of annotations we use a quarter of the validation set with ~40k images. Our results demonstrate that the proposed approach clearly outperforms direct fine-tuning across all major categories of classes in the Open Image dataset. Further, our approach is particularly effective for a large number of classes with wide range of noise in annotations (20-80% false positive annotations).

Citations

• Distilling the Knowledge in a Neural Network
• Training Convolutional Networks with Noisy Labels
• Gradient-based learning applied to document recognition
• Do Deep Nets Really Need to be Deep?
• Webly Supervised Learning of Convolutional Networks
• Unsupervised Learning of Visual Representations using Videos
• CNN Features off-the-shelf: an Astounding Baseline for Recognition

Cited by

• ErrorEraser: Unlearning Data Bias for Improved Continual Learning
• Distributionally Robust Optimization via Diffusion Ambiguity Modeling
• Deep Neural Networks Inspired by Differential Equations
• When Big Models Train Small Ones: Label-Free Model Parity Alignment for Efficient Visual Question Answering using Small VLMs
• Least-Ambiguous Multi-Label Classifier
• DART: Dual Adaptive Refinement Transfer for Open-Vocabulary Multi-Label Recognition
• Deep Self-Learning From Noisy Labels
• Derivative Manipulation for General Example Weighting
• Learning Category Correlations for Multi-label Image Recognition with Graph Networks
• Iterative Learning with Open-set Noisy Labels

Related

• Self-training with Noisy Student improves ImageNet classification
• Improving Training on Noisy Stuctured Labels
• CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise
• Image recognition from raw labels collected without annotators
• Big Self-Supervised Models are Strong Semi-Supervised Learners