Time-Warping Recurrent Neural Networks for Transfer Learning
This work addresses transfer learning challenges in dynamical systems modeling, particularly for wildfire prediction, but is incremental as it builds on existing time-warping concepts and applies them to RNNs.
The paper tackles the problem of transfer learning for Recurrent Neural Networks (RNNs) by proposing a time-warping method, which rescales time in models of physical systems, and demonstrates its effectiveness in predicting fuel moisture content for wildfire modeling with accuracy comparable to established methods while modifying fewer parameters.
Dynamical systems describe how a physical system evolves over time. Physical processes can evolve faster or slower in different environmental conditions. We use time-warping as rescaling the time in a model of a physical system. This thesis proposes a new method of transfer learning for Recurrent Neural Networks (RNNs) based on time-warping. We prove that for a class of linear, first-order differential equations known as time lag models, an LSTM can approximate these systems with any desired accuracy, and the model can be time-warped while maintaining the approximation accuracy. The Time-Warping method of transfer learning is then evaluated in an applied problem on predicting fuel moisture content (FMC), an important concept in wildfire modeling. An RNN with LSTM recurrent layers is pretrained on fuels with a characteristic time scale of 10 hours, where there are large quantities of data available for training. The RNN is then modified with transfer learning to generate predictions for fuels with characteristic time scales of 1 hour, 100 hours, and 1000 hours. The Time-Warping method is evaluated against several known methods of transfer learning. The Time-Warping method produces predictions with an accuracy level comparable to the established methods, despite modifying only a small fraction of the parameters that the other methods modify.