IEEE ISIT 2020

Abstract

It is highly likely that machine learning will play a key role in the next generation technology standard for cellular networks: 5G+ML=6G. In this talk I will review a number of ways I have been involved in exploring and pushing the boundary of that technology, based on a principle which I call Neural Augmentation (NA). In NA we acknowledge that the classical solutions that have been developed in the signal processing community are an extremely strong baseline on which we should build. On the other hand, we know that deep learning is able to learn patterns that are difficult or impossible to detect by humans, if it has access to a sufficiently large dataset and when the domain is narrow enough so that the acquired data can cover it. We argue that by combining classical engineering solutions with deep learning we can learn from smaller datasets and generalize better to out-of-domain situations. In NA we train a neural network to iteratively correct the classical solution. These corrections are hopefully small, and therefore more easy to model. We apply this principle to three problems in wireless communication: error-correction decoding, MIMO demodulation and channel estimation. We find that neural networks are indeed able to improve the state of the art when combined with the classical methods.

Abstract

The Poisson process and point processes in general have much to bring to the toolbox of the information theorist, apart from arising naturally when building models in some information theoretic problems of physical interest. This talk will attempt to reel up some of the times point processes have bitten the hook of researchers in the river of information theory.

The aim is to highlight some striking results and some intriguing techniques in this area.

Abstract

String reconstruction problems frequently arise in genomic data processing, molecular storage system implementations and synthetic biology. In the most general setting, the problems may be summarized as follows: one is given a single or multiple copies of a string. The copies are subsequently processed and transmitted through noise-inducing channels. The goal of the reconstruction method is to recover the original string or an approximation thereof using the noisy string information. Examples of string reconstruction questions include reconstruction from noisy traces, reconstruction from substrings and k-decks and reconstruction from substring composition sets. We review the above and related problems and then proceed to describe coding methods that lead to collections of strings that can be more accurately and efficiently reconstructed than their uncoded counterparts.

Abstract

Over the past two decades, semidefinite programming and sum of squares methods have provided state-of-the-art results — both theoretical and practical -- for a variety of problems in many areas, including combinatorial optimization, systems and control, and statistical estimation. In this talk we’ll provide a gentle introduction and survey of basic notions, algorithmic techniques, and future challenges. Particular emphasis will be given to newer developments, and applications to probability theory and quantum information.