One major challenge in training Deep Neural Networks is preventing overfitting. Many techniques such as data augmentation and novel regularizers such as Dropout have been proposed to prevent overfitting without requiring a massive amount of training data. In this work, we propose a new regularizer called DeCov which leads to significantly reduced overfitting (as indicated by the difference betw...

We propose a stable and reduced-complexity detection method for the Bell Labs layered Space-Time (BLAST) coding system. The existing iterative nulling and cancellation algorithm for BLAST has high computational complexity and requires repeated matrix pseudoinverse calculation which may lead to numerical instability. A square-root algorithm was proposed by other researchers to reduce complexity ...

Identifying the quantum chromodynamics (QCD) color structure of processes provides additional information to enhance reach for new physics searches at large Hadron collider (LHC). Analyses QCD in decay process a boosted particle have been spotted as becomes well localized limited phase space. While these kinds jet analyses provide an efficient way identify structure, constrained space reduces n...

With the increased relevance of metasurface for optical applications, a fabrication method that allows both large surface and 100 nm range dimensions at low cost is required their development. Due to its high throughput small structuration capabilities, Soft Nanoimprint Lithography good canditate as these type devices. But application metasurfaces visible wavelengths has been hindered by necess...

This paper presents an overview of wavelet-based image coding. We develop the basics of image coding with a discussion of vector quantization. We motivate the use of transform coding in practical settings, and describe the properties of various decorrelating transforms. We motivate the use of the wavelet transform in coding using rate-distortion considerations as well as approximation-theoretic...

We have investigated decorrelation of samples in quantum Monte Carlo ground-state energy calculations for large Li and H2O nanoclusters. Binning data as a way of eliminating statistical correlations, as is the common practice, is found to become increasingly impractical as the system size grows. We demonstrate nevertheless that it is possible to perform accurate energy calculations—without deco...

Sea surface height anomalies measured by the TOPEX/Poseidon satellite altimeter indicate high values of skewness and kurtosis. Except in a few regions, including the Gulf Stream, the Kuroshio Extension, and the Agulhas Retroflection, which display bimodal patterns of sea surface height variability, kurtosis is uniformly greater than 1.5 times the squared skewness minus an adjustment constant. T...

Neural one-unit learning rules for the problem of Independent Component Analysis (ICA) and blind source separation are introduced. In these new algorithms, every ICA neuron develops into a separator that finds one of the independent components. The learning rules use very simple constrained Hebbianjanti-Hebbian learning in which decorrelating feedback may be added. To speed up the convergence o...

We show a simple method for increasing on-average capacity of correlated MIMO channels, given limited channel knowledge at the transmitter. We further show that under certain channel constraints, this capacity approaches the capacity gained by water-filling on the complete channel. We give results which determine the benefit of applying a decorrelating, or whitening, matrix to the transmit sign...