Details

Autor(en) / Beteiligte

• Książek, Kamil
• Głomb, Przemysław
• Romaszewski, Michał
• Cholewa, Michał
• Grabowski, Bartosz
• Búza, Krisztián

Titel

Improving Autoencoder Training Performance for Hyperspectral Unmixing with Network Reinitialisation

Ist Teil von

• Image Analysis and Processing – ICIAP 2022, p.391-403

Ort / Verlag

Cham: Springer International Publishing

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• 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.