A
Andreas Spanias
Researcher at Arizona State University
Publications - 512
Citations - 8918
Andreas Spanias is an academic researcher from Arizona State University. The author has contributed to research in topics: Speech coding & Speech processing. The author has an hindex of 36, co-authored 490 publications receiving 7895 citations. Previous affiliations of Andreas Spanias include Arizona's Public Universities & Intel.
Papers
More filters
Proceedings ArticleDOI
A two stage approach for adaptive prediction of ARMA processes
TL;DR: A two-stage predictor is proposed, capable of predicting ARMA processes accurately with a reduced number of parameters by cascading a classical linear predictor, where the constraint on the predictor order is relaxed, and a pole-zero recursive like structure.
Posted Content
On the Design of Deep Priors for Unsupervised Audio Restoration
TL;DR: In this article, a new U-Net based audio prior architecture is proposed that does not impact either the network complexity or convergence behavior of existing convolutional architectures, yet leads to significantly improved audio restoration.
Proceedings ArticleDOI
Spatially-Varying Sharpness Map Estimation Based on the Quotient of Spectral Bands
TL;DR: A sharpness metric based on the quotient of high- to low-frequency bands of the log-spectrum of the image gradients is proposed and a descriptive dense sharpness map is obtained.
Proceedings ArticleDOI
Predicting the Generalization Gap in Deep Models using Anchoring
Vivek Sivaraman Narayanaswamy,R Anirudh,Irene Kim,Yamen Mubarka,Andreas Spanias,Jayaraman J. Thiagarajan +5 more
TL;DR: This paper proposes a novel strategy for directly predicting accuracy on unseen target data with the help of anchoring and pre-text encoding in predictive models and indicates that this approach produces well calibrated accuracy estimates outperforming existing baselines.
Proceedings ArticleDOI
Topology of high-contrast patches in SAR images
TL;DR: This analysis is extended to log-magnitude SAR images from the MSTAR database and shows that the most representative high contrast patches in SAR images lie among the clutter however methods extracting target patches only show results more similar to that obtained for natural imagery.