Abstract
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.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Change history
04 August 2022
In the originally published version of chapter 33, table 2 included an error. This has been corrected.
Notes
- 1.
With [12] method using uniform initialisation we have used the version from the PyTorch library, which differs from the paper with: 1) biases are not initialised to 0; 2) bounds of a uniform distribution are constant and not dependent on the number of connections.
References
Alabdulmohsin, I., Maennel, H., Keysers, D.: The impact of reinitialization on generalization in convolutional neural networks (2021)
Bingham, G., Miikkulainen, R.: AutoInit: analytic signal-preserving weight initialization for neural networks (2021)
Bioucas-Dias, J.M.: A variable splitting augmented Lagrangian approach to linear spectral unmixing. In: 2009 First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, pp. 1–4 (2009). https://doi.org/10.1109/WHISPERS.2009.5289072
Hyperspectral unmixing overview: geometrical, statistical, and sparse regression-based approaches. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 5(2), 354–379 (2012). https://doi.org/10.1109/JSTARS.2012.2194696
Boardman, J., Kruse, F.A., Green, R.: Mapping target signatures via partial unmixing of AVIRIS data. In: Summaries of the Fifth Annual JPL Airborne Earth Science Workshop. Volume 1: AVIRIS Workshop (1995)
Borsoi, R.A., Imbiriba, T., Bermudez, J.C.M.: Deep generative endmember modeling: an application to unsupervised spectral unmixing. IEEE Trans. Comput. Imaging 6, 374–384 (2020). https://doi.org/10.1109/TCI.2019.2948726
Conover, W.J.: Practical Nonparametric Statistics, vol. 350, 3rd edn. Wiley, Hoboken (1998)
Conover, W.J., Iman, R.L.: On multiple-comparisons procedures (1979)
Glorot, X., Bengio, Y.: Understanding the difficulty of training deep feedforward neural networks. In: Proceedings of AISTATS (2010), vol. 9, pp. 249–256 (2010)
Guo, A.J., Zhu, F.: Improving deep hyperspectral image classification performance with spectral unmixing. Signal Process. 183, 107949 (2021). https://doi.org/10.1016/j.sigpro.2020.107949
Guo, R., Wang, W., Qi, H.: Hyperspectral image unmixing using autoencoder cascade. In: 2015 7th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), pp. 1–4 (2015). https://doi.org/10.1109/WHISPERS.2015.8075378
He, K., Zhang, X., Ren, S., Sun, J.: Delving deep into rectifiers: surpassing human-level performance on ImageNet classification. In: Proceedings of ICCV, pp. 1026–1034 (2015). https://doi.org/10.1109/ICCV.2015.123
Hinton, G.E., Salakhutdinov, R.R.: Reducing the dimensionality of data with neural networks. Science 313(5786), 504–507 (2006). https://doi.org/10.1126/science.1127647
Keshava, N., Mustard, J.F.: Spectral unmixing. IEEE Signal Process. Mag. 19(1), 44–57 (2002). https://doi.org/10.1109/79.974727
Krizhevsky, A., Hinton, G.E.: Using very deep autoencoders for content-based image retrieval. In: ESANN (2011)
Kruskal, W.H., Wallis, W.A.: Use of ranks in one-criterion variance analysis. J. Am. Stat. Assoc. 47(260), 583–621 (1952). https://doi.org/10.2307/2280779
LeCun, Y.A., Bottou, L., Orr, G.B., Müller, K.-R.: Efficient BackProp. In: Montavon, G., Orr, G.B., Müller, K.-R. (eds.) Neural Networks: Tricks of the Trade. LNCS, vol. 7700, pp. 9–48. Springer, Heidelberg (2012). https://doi.org/10.1007/978-3-642-35289-8_3
Liaw, R., Liang, E., Nishihara, R., Moritz, P., Gonzalez, J.E., Stoica, I.: Tune: a research platform for distributed model selection and training (2018)
Lu, L., Shin, Y., Su, Y., Em Karniadakis, G.: Dying ReLU and initialization: theory and numerical examples. Commun. Comput. Phys. 28(5), 1671–1706 (2020). https://doi.org/10.4208/cicp.OA-2020-0165
Lv, J., Shao, X., Xing, J., Cheng, C., Zhou, X.: A deep regression architecture with two-stage re-initialization for high performance facial landmark detection. In: Proceedings of CVPR 2017, pp. 3691–3700 (2017). https://doi.org/10.1109/CVPR.2017.393
Ozkan, S., Kaya, B., Akar, G.B.: EndNet: sparse autoencoder network for endmember extraction and hyperspectral unmixing. IEEE Trans. Geosci. Remote Sens. 57(1), 482–496 (2019). https://doi.org/10.1109/TGRS.2018.2856929
Palsson, B., Ulfarsson, M.O., Sveinsson, J.R.: Convolutional autoencoder for spatial-spectral hyperspectral unmixing. In: Proceedings of IGARSS 2019, pp. 357–360 (2019). https://doi.org/10.1109/IGARSS.2019.8900297
Palsson, B., Sigurdsson, J., Sveinsson, J.R., Ulfarsson, M.O.: Hyperspectral unmixing using a neural network autoencoder. IEEE Access 6, 25646–25656 (2018). https://doi.org/10.1109/ACCESS.2018.2818280
Palsson, B., Ulfarsson, M.O., Sveinsson, J.R.: Convolutional autoencoder for spectral-spatial hyperspectral unmixing. IEEE Trans. Geosci. Remote Sens. 59(1), 535–549 (2021). https://doi.org/10.1109/IGARSS.2019.8900297
Plaza, A., Chang, C.: Impact of initialization on design of endmember extraction algorithms. IEEE Trans. Geosci. Remote Sens. 44(11), 3397–3407 (2006). https://doi.org/10.1109/TGRS.2006.879538
Ranasinghe, Y., et al.: Convolutional autoencoder for blind hyperspectral image unmixing (2020)
Rister, B., Rubin, D.L.: Probabilistic bounds on neuron death in deep rectifier networks (2021)
Su, Y., Li, J., Plaza, A., Marinoni, A., Gamba, P., Chakravortty, S.: DAEN: deep autoencoder networks for hyperspectral unmixing. IEEE Trans. Geosci. Remote Sens. 57(7), 4309–4321 (2019). https://doi.org/10.1109/TGRS.2018.2890633
Winter, M.E.: N-FINDR: an algorithm for fast autonomous spectral end-member determination in hyperspectral data. In: Descour, M.R., Shen, S.S. (eds.) Imaging Spectrometry V, vol. 3753, pp. 266–275. International Society for Optics and Photonics, SPIE (1999). https://doi.org/10.1117/12.366289
Zhao, M., Wang, M., Chen, J., Rahardja, S.: Hyperspectral unmixing via deep autoencoder networks for a generalized linear-mixture/nonlinear-fluctuation model (2019)
Zhu, F.: Hyperspectral unmixing: ground truth labeling, datasets, benchmark performances and survey (2017)
Acknowledgements
K.K. acknowledges funding from the European Union through the European Social Fund (grant POWR.03.02.00-00-I029). B.G. acknowledges funding from the budget funds for science in the years 2018-2022, as a scientific project “Application of transfer learning methods in the problem of hyperspectral images classification using convolutional neural networks” under the “Diamond Grant” program, no. DI2017 013847.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
1 Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Książek, K., Głomb, P., Romaszewski, M., Cholewa, M., Grabowski, B., Búza, K. (2022). Improving Autoencoder Training Performance for Hyperspectral Unmixing with Network Reinitialisation. In: Sclaroff, S., Distante, C., Leo, M., Farinella, G.M., Tombari, F. (eds) Image Analysis and Processing – ICIAP 2022. ICIAP 2022. Lecture Notes in Computer Science, vol 13231. Springer, Cham. https://doi.org/10.1007/978-3-031-06427-2_33
Download citation
DOI: https://doi.org/10.1007/978-3-031-06427-2_33
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-06426-5
Online ISBN: 978-3-031-06427-2
eBook Packages: Computer ScienceComputer Science (R0)