Our recent paper titled “Physics-Guided Loss Functions Improve Deep Learning Performance in Inverse Scattering” has been recognized for its significant impact and popularity in the scientific community. It has earned the distinction of being one of the top 25 downloaded articles between September 2022 and September 2023 in the IEEE Transactions on Computational Imaging, as reported by IEEE Xplore®. This achievement is a testament to the hard work and innovative thinking of the authors: Zicheng Liu, Mayank Roy, and Krishna Agarwal.
The paper delves into the complex realm of electromagnetic inverse scattering problems (ISPs), which have long been a challenge due to their nonlinearity, ill-posedness, and high computational demands. By leveraging deep neural network (DNN) techniques, our team has demonstrated the potential to surpass traditional imaging methods. The crux of our research lies in the novel approach of integrating physical principles directly into the training of DNNs, specifically by embedding near-field priors into the learning process.
This recognition from IEEE Xplore® is not only an honor but also an encouragement for our team to continue exploring the frontiers of computational imaging and deep learning. We extend our deepest thanks to the readers and peers who have engaged with our work, and we look forward to contributing further to the scientific dialogue in this exciting field.
A special shoutout to Zicheng, Mayank, and Krishna for their exceptional contributions.