Benchmarking Invertible Architectures on Inverse Problems
This work provides an initial evaluation framework for invertible architectures on inverse problems, aiming to inspire community benchmarks.
The researchers benchmarked ten invertible architectures on two low-dimensional inverse problems, finding that coupling layers and simple autoencoders achieved the best results.
Recent work demonstrated that flow-based invertible neural networks are promising tools for solving ambiguous inverse problems. Following up on this, we investigate how ten invertible architectures and related models fare on two intuitive, low-dimensional benchmark problems, obtaining the best results with coupling layers and simple autoencoders. We hope that our initial efforts inspire other researchers to evaluate their invertible architectures in the same setting and put forth additional benchmarks, so our evaluation may eventually grow into an official community challenge.