# Diffusion-based Deep Active Learning

arXiv: https://arxiv.org/abs/2003.10339 · HTML: https://mlsanity.com/paper/2003.10339/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 23 Mar 2020.

## Built on by (most cited first, 10 of 10)

- [GALAXY: Graph-based Active Learning at the Extreme](https://mlsanity.com/paper/2202.01402/) (2022)
- [Model-Change Active Learning in Graph-Based Semi-Supervised Learning](https://mlsanity.com/paper/2110.07739/) (2021)
- [Poisson Reweighted Laplacian Uncertainty Sampling for Graph-based Active Learning](https://mlsanity.com/paper/2210.15786/) (2022)
- [Diffusion Mechanism in Residual Neural Network: Theory and Applications](https://mlsanity.com/paper/2105.03155/) (2021)
- [Maximizing and Satisficing in Multi-armed Bandits with Graph Information](https://mlsanity.com/paper/2108.01152/) (2021)
- [Novel Batch Active Learning Approach and Its Application to Synthetic Aperture Radar Datasets](https://mlsanity.com/paper/2307.10495/) (2023)
- [MALADY: Multiclass Active Learning with Auction Dynamics on Graphs](https://mlsanity.com/paper/2409.09475/) (2024)
- [An efficient scheme based on graph centrality to select nodes for training for effective learning](https://mlsanity.com/paper/2104.14123/) (2021)
- [Hard to See, Hard to Label: Generative and Symbolic Acquisition for Subtle Visual Phenomena](https://mlsanity.com/paper/2604.22990/) (2026)
- [An Approximation Algorithm for Graph Label Selection](https://mlsanity.com/paper/2605.18623/) (2026)

## Abstract (arXiv)

The remarkable performance of deep neural networks depends on the availability of massive labeled data. To alleviate the load of data annotation, active deep learning aims to select a minimal set of training points to be labelled which yields maximal model accuracy. Most existing approaches implement either an `exploration'-type selection criterion, which aims at exploring the joint distribution of data and labels, or a `refinement'-type criterion which aims at localizing the detected decision boundaries. We propose a versatile and efficient criterion that automatically switches from exploration to refinement when the distribution has been sufficiently mapped. Our criterion relies on a process of diffusing the existing label information over a graph constructed from the hidden representation of the data set as provided by the neural network. This graph representation captures the intrinsic geometry of the approximated labeling function. The diffusion-based criterion is shown to be advantageous as it outperforms existing criteria for deep active learning.