Deep Individual Active Learning: Safeguarding against Out-of-Distribution Challenges in Neural Networks.

Active learning (AL) is a paradigm focused on purposefully selecting training data to enhance a model's performance by minimizing the need for annotated samples. Typically, strategies assume that the training pool shares the same distribution as the test set, which is not always valid in privacy-sensitive applications where annotating user data is challenging. In this study, we operate within an individual setting and leverage an active learning criterion which selects data points for labeling based on minimizing the min-max regret on a small unlabeled test set sample.

Robust Universal Inference.

Learning and making inference from a finite set of samples are among the fundamental problems in science. In most popular applications, the paradigmatic approach is to seek a model that best explains the data. This approach has many desirable properties when the number of samples is large.

Nonlinear Canonical Correlation Analysis:A Compressed Representation Approach.

Canonical Correlation Analysis (CCA) is a linear representation learning method that seeks maximally correlated variables in multi-view data. Nonlinear CCA extends this notion to a broader family of transformations, which are more powerful in many real-world applications. Given the joint probability, the Alternating Conditional Expectation (ACE) algorithm provides an optimal solution to the nonlinear CCA problem.

Accumulation of nonlinear interference noise in fiber-optic systems.

Through a series of extensive system simulations we show that all of the previously not understood discrepancies between the Gaussian noise (GN) model and simulations can be attributed to the omission of an important, recently reported, fourth-order noise (FON) term, that accounts for the statistical dependencies within the spectrum of the interfering channel. We examine the importance of the FON term as well as the dependence of NLIN on modulation format with respect to link-length and number of spans. A computationally efficient method for evaluating the FON contribution, as well as the overall NLIN power is provided.

New bounds on the capacity of the nonlinear fiber-optic channel.

We revisit the problem of estimating the nonlinear channel capacity of fiber-optic systems. By taking advantage of the fact that a large fraction of the nonlinear interference between different wavelength-division-multiplexed channels manifests itself as phase noise, and by accounting for the long temporal correlations of this noise, we show that the capacity is notably higher than what is currently assumed. This advantage translates into nearly doubling of the link distance for a fixed transmission rate.

Properties of nonlinear noise in long, dispersion-uncompensated fiber links.

We study the properties of nonlinear interference noise (NLIN) in fiber-optic communications systems with large accumulated dispersion. Our focus is on settling the discrepancy between the results of the Gaussian noise (GN) model (according to which NLIN is additive Gaussian) and a recently published time-domain analysis, which attributes drastically different properties to the NLIN. Upon reviewing the two approaches we identify several unjustified assumptions that are key in the derivation of the GN model, and that are responsible for the discrepancy.

Optical implementation of a space-time-trellis code for enhancing the tolerance of systems to polarization-dependent loss.

We propose a space-time coding scheme designed to increase the tolerance of fiber-optic communications systems to polarization-dependent loss (PDL). A notable increase in the tolerable amount of average link PDL is achieved without affecting the complexity of the overall optical communications link. Other advantages include seamless integration with the broadly deployed blind equalization modules relying on the constant modulus algorithm.

Beneficial use of spectral broadening resulting from the nonlinearity of the fiber-optic channel.

We discuss the possibility of exploiting spectral broadening resulting from fiber nonlinearity for the transmission of information. The spectral broadening induced by nonlinearity combined with the appropriate waveform can turn quadrature amplitude modulation-like constellations into frequency-shift-keying constellations over a much larger dimension. Thus, the Kerr effect can be thought of as a large dimensional mapper/modulator.

The underaddressed optical multiple-input, multiple-output channel: capacity and outage.

We study an optical space-division multiplexed system where the number of modes that are addressed by the transmitter and receiver is allowed to be smaller than the total number of optical modes supported by the fiber. This situation will be of relevance if, for instance, fibers supporting more modes than can be processed with current multiple-input, multiple-output technology are deployed with the purpose of future-proof installation. We calculate the ergodic capacity and the outage probability of the link and study their dependence on the number of addressed modes at the transmitter and receiver.

Use of space-time coding in coherent polarization-multiplexed systems suffering from polarization-dependent loss.

We evaluate the advantage of using space-time coding in order to increase the tolerance of fiber-optic communications systems to polarization-dependent loss (PDL). Focusing on three particular codes, the Golden Code, the Silver Code (SC), and the Alamouti Code (AC), we calculate the amount of average PDL that can be tolerated for a given signal-to-noise ratio margin that is designed into the system. The SC is shown to be optimal in the case of low to moderate PDL, whereas, in the case of extreme PDL, the AC shows the best performance.