Several applications of Internet of Things (IoT) technology involve capturing data from multiple sensors resulting in multi-sensor time series . Such approaches can struggle in the practical setting where different instances of the same device or equipment such as mobiles, wearables, engines, etc. come with different combinations of installed sensors . We propose a novel neural network architecture suitable for zero-shot transfer learning allowing robust inference for multivariate time series with previously unseen combination of available dimensions or sensors at test time . The proposed approach allows for better generalization in comparison to a deep gated recurrent neural network baseline . We evaluate the proposed approach on publicly available activity recognition and equipment prognostics datasets, and show that the proposed approaches allow for better generalization in comparison to the proposed approval to a deep network baseline. We also show that proposed approach is effective and that the proposal allows for the proposed application of this type of approach to a neural network model to be used in our analysis of publicly available data sets and that it is effective in our new analysis of public data sets. We are happy to use this data sets to test our ability to identify and

Links: PDF - Abstract

Code :


Keywords : proposed - network - approach - time - data -

Leave a Reply

Your email address will not be published. Required fields are marked *


Enjoy this blog? Please spread the word :)