Computer Science
8 years ago

New method distinguishes complex systems with time-varying parameters effortlessly.

Advances in intelligent systems and computing
Bilinear Representation of Non-stationary Autoregressive Time Series
Ewa Bielińska

Paper Summary

Related papers

Theme background image
Engineering
2 years ago

New autoregressive models revolutionize forecasting for active energy complexes!

Electricity Forecasting
Stock Market Predictions
Oil Price Shocks
1 view
Decision Sciences
2 years ago

Revolutionary Models Predict Future Events with Unprecedented Accuracy

Time Series Forecasting
1 view
Theme background image
Computer Science
2 years ago

Non-autoregressive neural networks outperform autoregressive for faster system identification.

Neural Networks
Anomaly Detection
Image Denoising
1 view

The article discusses a way to deal with non-stationary autoregressive systems where parameters change over time. It can be hard to tell these systems apart just by looking at the output signal. The researchers suggest using a bilinear approximation to simplify the model and make it easier to distinguish between different systems. This method replaces time-varying parameters with constant ones, making the systems easier to compare.