Visualization and interpretation of deep learning models
Post date: Feb 10, 2018 8:36:32 PM
We visualise the LSTM deep learning models by principal component analysis. The similarity of the events in fault isolation can be evaluated based on the linear embedding layer of the network, which maps the input signals into a continuous-valued vector space. The method is demonstrated in a simulated vinyl acetate production technology. The results illustrate that with the application of RNN based sequence learning not only accurate fault classification solutions can be developed, but the visualisation of the model can give useful hints for hazard analysis.
The paper related paper will be published in Journal of Chemometrics soon.
The algorithm was implemented in Python. The related code can be downloaded from our Github repository.