# Improving Autoencoder Training Performance for Hyperspectral Unmixing with Network Reinitialisation

arXiv: https://arxiv.org/abs/2109.13748 · HTML: https://mlsanity.com/paper/2109.13748/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 33 references that are in the map.
- First posted on arXiv 28 Sep 2021.

## Builds on (most cited first, 9 of 9)

- [Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification](https://mlsanity.com/paper/1502.01852/) (2015)
- [Tune: A Research Platform for Distributed Model Selection and Training](https://mlsanity.com/paper/1807.05118/) (2018)
- [Hyperspectral Unmixing: Ground Truth Labeling, Datasets, Benchmark Performances and Survey](https://mlsanity.com/paper/1708.05125/) (2017)
- [EndNet: Sparse AutoEncoder Network for Endmember Extraction and Hyperspectral Unmixing](https://mlsanity.com/paper/1708.01894/) (2017)
- [The Impact of Reinitialization on Generalization in Convolutional Neural Networks](https://mlsanity.com/paper/2109.00267/) (2021)
- [Deep Generative Endmember Modeling: An Application to Unsupervised Spectral Unmixing](https://mlsanity.com/paper/1902.05528/) (2019)
- [AutoInit: Analytic Signal-Preserving Weight Initialization for Neural Networks](https://mlsanity.com/paper/2109.08958/) (2021)
- [Convolutional Autoencoder for Blind Hyperspectral Image Unmixing](https://mlsanity.com/paper/2011.09420/) (2020)
- [Probabilistic bounds on neuron death in deep rectifier networks](https://mlsanity.com/paper/2007.06192/) (2020)

## Built on by (most cited first, 1 of 1)

- [CRAFT: Contextual Re-Activation of Filters for face recognition Training](https://mlsanity.com/paper/2312.00072/) (2023)

## Abstract (arXiv)

Neural networks, in particular autoencoders, are one of the most promising solutions for unmixing hyperspectral data, i.e. reconstructing the spectra of observed substances (endmembers) and their relative mixing fractions (abundances), which is needed for effective hyperspectral analysis and classification. However, as we show in this paper, the training of autoencoders for unmixing is highly dependent on weights initialisation; some sets of weights lead to degenerate or low-performance solutions, introducing negative bias in the expected performance. In this work, we experimentally investigate autoencoders stability as well as network reinitialisation methods based on coefficients of neurons' dead activations. We demonstrate that the proposed techniques have a positive effect on autoencoder training in terms of reconstruction, abundances and endmembers errors.