Keynote Speakers

Lisheng Xu

Prof. Dr.
Professor of College of Medicine and Biological Information Engineering, Hunnan Campus of Northeastern University, China
Director of the biomedical electronics institute in Sino-Dutch Biomedical and Information Engineering School of Northeastern University;
Professor at the Key Laboratory of Medical Image Computing of Ministry of Education, China;
Chairman of theory and education professional committee of China Medical Informatics Association;
Vice chairman of TCM diagnosis information committee of Chinese medicine information society;
Senior member of IEEE Society;
Senior member of Chinese Society of Biomedical Engineering;
Member of the editor board for many international journals such as Physiological Measurement, Biomedical Engineering Online, Computers in Biology and Medicine and so on.

Title:Noninvasive monitoring of aortic pressure wave based on peripheral pulse waves

Abstract: Blood pressure wave can reflect the status of cardiovascular system. In comparison to peripheral arterial pressure waves, aortic pressure wave is more efficient in predicting cardiovascular events. However, the direct measurement of aortic pressure wave is invasive, complex, expensive, and has some high risks. Therefore, the non-invasive estimation of aortic pressure wave has attracted many concerns. The transfer function methods have been studied for a long time and have achieved many successes. Currently, the transfer function has changed from population-based method to individualized method. There are many ways to individualize the transfer function of cardiovascular system. This presentation will show some work of our group to individualize the transfer function by adaptively adjust the transfer function according to the systolic blood pressure acquired from the arm cuff, considering the time-variant characteristics of cardiovascular system using blind identification technique, combining with 0D (Windkessel) models, combining with temporal convolutional network. Furthermore, the advantages and disadvantages of these methods will also be discussed.

Teodor Bulboaca,

Babes-Bolyai University, Romania

Title: Special Functions and Their Remarkable Geometric Properties

In the first part of this lecture we establish geometric properties, such as starlikeness and convexity of order α (0 ≤ α < 1), and close-to-convexity in the open unit disk U of the complex plane for a combination of a normalized form of the generalized Struve function of order p, wp,b,c(z), defined by Dp,b,c(z) = 2p√ π Γ(p+b/2+1)z(−p+1)/2dp,b,c( √ z), where dp,b,c(z) := pwp,b,c(z)+zw′ p,b,c(z), with p, b, c ∈ C and κ = p + b/2 + 1 ∈/ {0, −1, −2, . . . }. We determine conditions for the parameters c and κ for which f ∈ R(β) = {f ∈ A(U) : Re f′(z) > β, z ∈ Up} (0 ≤ β < 1) indicates that the convolution product Dp,b,c ∗ f belongs to the spaces H∞(U) and R(γ) with γ depending on α and β, where A(U) denotes the class of all normalized analytic functions in U and H∞(U) is the space of all bounded analytic functions in A(U). We also obtain sufficient conditions in terms of the expansion coefficients for f ∈ A(U) to be in some subclasses of the class of univalent functions. Motivation has come from the vital role of special functions in Geometric Function Theory.

Sérgio Duarte Correia

Associate Professor
Polytechnic Institute of Portalegre, Technology and Management School, Portugal Head of the Electronics and Instrumentation Laboratory, Syllabus/Course Coordination, Teaching and Research, Vice-President of the Pedagogical Council

Title: Numerical Quantization of Neural Networks

Artificial Intelligence (AI) and its usage are becoming quite regular and spread daily. While applications such as chatbots, recommender systems, or spam filters rely on massive network structures that are deployed on remote servers, when it comes to natural language processing or healthcare monitoring, it is pretty crucial that AI runs its algorithms on a smartphone or a wearable device, both with memory and power restrictions. Also, running AI algorithms on edge or on the device can solve many issues due to privacy concerns or bandwidth limitations. This keynote will discuss the challenges in transforming numeric quantities to low-bit-width representation and the numerical error that arises.

Golubev Vasily

Associate Professor
Moscow Institute of Physics and Technology, Laboratory of Applied Numerical Geophysics

Title: Continuum Models of Fractured and Elastoviscoplastic Media

The seismic survey is a key technology for the oil and gas deposit exploration. Novel algorithms for seismic inverse problems are capable of handling heterogeneous geological media but require high quality results of direct problem modeling. The most difficult object in the forward modeling of the geological media is the fractured inclusion. Different approaches exist to describe its dynamic behavior, mostly limited to the linear contact conditions on crack boundaries.
Continuum models of layered and block media with the nonlinear slip at contact boundaries may be substantiated in the frame of the asymptotic averaging theory. In the zeroth-order approximation, they coincide with models constructed using slip theory relations. This work presents the nonlinear continuum model of the layered medium with visco-plastic interlayers and adopts it to the dynamic problem of wave propagation. The model relies on the linear isotropic theory and the theory of periodic media. The slip velocity vector and the delamination velocity vector are treated as continuous functions of time and coordinates.
This presentation will show the application of the continuum models to the applied problems of the oil deposit exploration and nondestructive control of composite materials.