One-dimensional Deep Image Prior for Time Series Inverse Problems

We extend the Deep Image Prior (DIP) framework to one-dimensional time series signals. DIP is using a randomly initialized convolutional neural network (CNN) to solve linear inverse problems by optimizing over the weights of the network to fit the observed measurements. Our main finding is that properly tuned one-dimensional convolutional architectures provide an excellent Deep Prior for various types of temporal signals including audio, biological signals, and sensor measurements. We show that our network can be used for a wide range of inverse tasks including missing value imputation, forecasting, and blind denoising. Our method outperforms baselines even with a fraction of the observed measurements in a variety of recovery tasks.