The role of consumer-grade wearable devices in disease prevention and diagnosis
The use of consumer-grade wearable devices has surged in the recent years. While fitness trackers and smart watches aim at improving fitness and wellbeing, more elaborate sensor systems embedded in clothing can sense a multitude of biosignals. They are therefore a continuous source of information on the physical status of oneself. Until now, only limited use has been made on this information, as data quality and interpretability are a problem for healthcare systems working at their capacity. In this lecture, we will investigate how smart wearable devices can aid patients and doctors before, during and after diagnosis and treatment and what is needed for a seamless integration into the healthcare enterprise.
Statistical limits of machine learning
Machine learning currently suffers from the ‘burden of stardom’ – people expect that every problem can be solved through data analysis. Machine learning algorithms are often applied to complex problems where it is hard to judge the performance of the algorithms, and hence, the fidelity of the solutions. The aim of learning theory is to mathematically understand when one can solve machine learning problems, and which algorithms are best suited for the job. In this talk, Prof. Ghoshdastidar will provide a glimpse to the statistical challenges of learning from large (high-dimensional) data. In particular, he will discuss the problem of hypothesis testing for large graphs and demonstrate that a conventional approach can result in ‘unsolvable’ statistical problems. However, when the questions are posed appropriately, one can develop methods with performance guarantees. He will present some applications of these methods in testing communication and biological networks.
Micro- to mesoscale simulation of strongly correlated quantum systems
We will briefly review some of the core concepts of quantum mechanics as well as one of the most intriguing and fascinating consequences, namely nonlocality and the Bell inequalities. The second part of the talk is concerned with modern computational methods, like the matrix product state framework and quantum Monte Carlo methods, to simulate (strongly correlated) quantum systems on classical computers. While the fundamental physical laws describing quantum systems are in principle well-known, the collective dynamical behavior of many-body quantum systems can lead to qualitatively new effects, a prominent example being superconductivity. Finally, we will look at the future roadmap for prospective quantum computing and quantum information technology.