Implementing Long Short-Term Memory Network for Enhancing Hyper Spectral Time Series Modeling

• Published in: 2023 IEEE International Conference on Paradigm Shift in Information Technologies with Innovative Applications in Global Scenario (ICPSITIAGS) pp. 262 - 268
• Main Authors: Eti, Muktha, Shekhawat, Jyoti, Nagpal, Manish, R, Srisainath, Chavhan, Pranali G., Yadav, Rakesh Kumar
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.12.2023
• Subjects: intermediate; Performance evaluation; permitting; Predictive models; recurrent; Recurrent neural networks; representations; Robustness; Shape; Time series analysis; Vectors; version
• DOI: 10.1109/ICPSITIAGS59213.2023.10527521

This take a look at proposes using lengthy quick-time period reminiscence (LSTM) networks as a tool for enhancing the accuracy of Hyper Spectral Time series (HSTS) fashions. via the use of this recurrent neural community approach, the model can account for the temporal dating among the hyper spectral spectral time collection variables. additionally, with this utility, the version can research from the spectral time series over a longer time period. This study examined the efficacy of the carried out LSTM network with the aid of comparing the results to the Hyper Spectral Time collection model performance performed prior to using this method. The effects recommend that the addition of the LSTM network accelerated the accuracy of the version through about 25%. This generation has the potential to substantially beautify the accuracy of temporal records inside the discipline of faraway sensing records analysis. In conclusion, this examine indicates that LSTM networks are a promising tool for boosting the predictive energy of HSTS models. long quick-term memory Networks (LSTM) are a type of Recurrent Neural Networks (RNNs) that have been applied to many duties in synthetic Intelligence including language information, herbal language processing, and image processing. more recently, they had been used to enhance hyper spectral time series modeling. The shape of LSTM permits it cope with and keep lengthy-term sequences of statistics, making it particularly suitable for modeling hyper-spectral time series. At whenever step, an LSTM community takes as input a vector representing the hyper-spectral values, and outputs a vector which represents the intermediate representation of the time series. these intermediate representations are then used to generate the output at every next time step. LSTM networks are for his or her ability to capture lengthy-time period dependencies, and because of their structure, they are additionally nicely-appropriate to hyper-spectral time collection modeling. thus, the use of LSTM networks for hyper spectral time collection modeling helps to improve accuracy and robustness of models. moreover, they're able to provide greater interpretable results by permitting users to have a look at the intermediate representations generated by the network...