Seminars

VU Institute of Computer science seminars are held every second wednesday at 4 pm.

Speakers and talks

General informatioon about seminars.

Due to pandemic seminars are held remotely.

Those wishing to read a report at the seminar are kindly invited to contact us by e-mail

mindaugas.bloznelis@mif.vu.lt

Seminarai 2020/2021 metais

2020 12 02. The speaker was prof. Tadas Meškauskas.

Report topic: Computer modeling of defective biological membranes

Summary: I will tell about the research carried out together with VU GMC colleagues (Prof. G.Valčius and his group) and II doctoral students T.Raila, S.Bucka. Field of application – electrochemical properties of toxin-damaged (defective) phospholipid membranes. These membranes mimic the cell membranes of living organisms, where exposure to toxins, inflammation, aging, and other causes can lead to defects that distort electrochemical transformations (which can be associated with neurodegenerative diseases: e.g., Alzheimer’s disease, Parkinson’s disease). Computer models of three-dimensional defective membranes (visually similar to pieces of Dutch cheese) are created and analyzed (by the finite element method).

How to use a computer to model the electrochemical response – impedance spectrum (measured and experimentally in the laboratory), what challenges are encountered? The inverse problem – having an impedance spectrum, what can be said about the type of membrane damage (the nature of defect clustering)? What metrics (ideas related to Voronoi diagrams) can be useful in quantifying the degree of defect clustering? Generation of synthetic data and their correspondence to real membrane images (obtained by atomic force microscopy) – how to pseudo-randomly scatter defects (ideas from computer graphics: simulations of clouds, smoke textures) with the nature of clustering observed in experiments? Challenges in automated defect recognition (in atomic force photographs) – “simple” (Hough transformation, etc.) and “heavy” (artificial intelligence) methodologies.

Literature:

1. T.Raila, F.Ambrulevičius, T.Penkauskas, M.Jankunec, T.Meškauskas, David J. Vanderah, G.Valincius, Clusters of protein pores in phospholipid bilayer membranes can be identified and characterized by electrochemical impedance spectroscopy, Electrochimica Acta, 364, 2020. https://doi.org/10.1016/j.electacta.2020.137179

2. T.Raila, T.Penkauskas, M.Jankunec, G.Dreižas, T.Meškauskas, G.Valincius, Electrochemical impedance of randomly distributed defects in tethered phospholipid bilayers: Finite element analysis, Electrochimica Acta, 299, pp. 863–874, 2019. https://doi.org/10.1016/j.electacta.2018.12.148

3. G.Valincius, T.Meškauskas, F.Ivanauskas, Electrochemical Impedance Spectroscopy of Tethered Bilayer Membranes, Langmuir, 28, pp. 977–990, 2012. http://dx.doi.org/10.1021/la204054g

4. T.Raila, M.Jankunec, T.Meškauskas, G.Valinčius, Computational models of defect clustering for tethered bilayer membranes, In: Proceedings of 20th International Conference ICCSA 2020 (Eds. O.Gervasi et al.), Springer, Lecture Notes in Computer Science, 12253, pp. 496–504, 2020. https://doi.org/10.1007/978-3-030-58814-4_35

5. T.Raila, T.Meškauskas, G.Valinčius, M.Jankunec, T.Penkauskas, Computer modeling of electrochemical impedance spectra for defected phospholipid membranes: finite element analysis, In: Proceedings of 3rd International Conference NUMTA 2019 (Eds. Y.D.Sergeyev and D.E.Kvasov), Springer, Lecture Notes in Computer Science, 11974, pp. 462–469, 2020. https://doi.org/10.1007/978-3-030-40616-5_44

November 18, 2020 Speaker prof. Leonidas Sakalauskas

Report topic: Mathematical-computer of social behavior phenomena
modeling

(The language of the lecture is Lithuanian.)

Summary: The paper is devoted to the applications of multi-agent modeling, gambling theory and structural probabilistic modeling in the systematic analysis of various social behavior phenomena (paradigms of group decisions of gambling theory, social capital dynamics, information aspects of pandemics-infodemias, etc.). The issues of ontological operationalization of real processes are analyzed in order to show the connections between computer science and social sciences and humanities, and to develop a methodology for the formation of Evidence-Based Decisions.

The report is based on the results of the LMT research group project “Development of Metrics, Conceptual and Simulation Model of Social Impact of Cultural Processes”, http://www.sicp.mii.vu.lt.

Literature:

1. E. Vilkas (2002) Sprendimų priėmimo teorija. VDU
2. R. Axelrod (2006), The Evolution of Cooperation.
3. J. Scott (2017). Social Network Analysis
4. K. J. Arrow (1963). Social Choice and Individual Values
5. I. Curriel (1997) Cooperative Game Theory and Applications
6. R. Bagozzi, Yi Youjae (2011). Specification, evaluation, and
interpretation of structural equation models. Journal of the Academy
of Marketing Science, vol 40, No 1, pp. 8–34.
7. L. Sakalauskas, V. Dulskis, R. Lauzikas, A. Miliauskas, D.
Plikynas (2020) A probabilistic model of the impact of cultural
participation on social capital. JOURNAL OF MATHEMATICAL SOCIOLOGY.
DOI: 10.1080/0022250X.2020.1725002
8. R. Karbauskaite, L. Sakalauskas, G. Dzemyda (2020) Kriging
Predictor for Facial Emotion Recognition Using Numerical Proximities
of Human Emotions. INFORMATICA, vol 31, No 2, pp. 249-275

November 4, 2020 Speaker dr. Linas Petkevičius.

Report topic: Image comparison measures for medical image reconstruction

Summary: Over the past five years, image analysis and deep learning methods have evolved particularly rapidly. Success in computer vision has enabled a variety of new applications including medical image analysis. Video reconstruction, noise cancellation, and other similar applications have evolved rapidly.
The workshop presents an overview and practical research on how image comparison measures affect the final accuracy of models. Research shows that the complex structure of images requires a very detailed selection of image comparison measures to create models that work in practice.

Literature:

1. Wang, Zhou, et al. “Image quality assessment: from error visibility to structural similarity.” IEEE transactions on image processing 13.4 (2004): 600-612.

2. You, Chenyu, et al. “Structurally-sensitive multi-scale deep neural network for low-dose CT denoising.” IEEE Access 6 (2018): 41839-41855.

3. Renieblas, Gabriel Prieto, et al. “Structural similarity index family for image quality assessment in radiological images.” Journal of medical imaging 4.3 (2017): 035501.1

4. Brusokas, Jonas; Petkevičius, Linas. Analysis of deep neural network architectures and similarity metrics for low-dose CT reconstruction // 2020 IEEE open conference of electrical, electronic and information sciences (eStream)), 30 April 2020, Vilnius, Lithuania: proceedings of the conference / organized by Vilnius Gediminas Technical University. New York : IEEE, 2020. ISBN 9781728197791. p. [1-6]. DOI: 10.1109/eStream50540.2020.9108883.

5. Brusokas, Jonas; Petkevičius, Linas. Numerical analysis of SLSSIM similarity on medical X-ray image domain // Informacinės technologijos = Information technology : XXIV tarpuniversitetinės tarptautinės magistrantų ir doktorantų konferencijos „Informacinė visuomenė ir universitetinės studijos“ (IVUS 2019) medžiaga, 2019 m. balandžio 25 d.,Kaunas,Lietuva. Kaunas : Vytauto Didžiojo universiteto leidykla. ISSN 2029-249X. eISSN 2029-4824. 2019, p. 119-124.

October 21, 2020 Speaker prof. Linas Laibinis

Report topic: The refinement calculus – a mathematical theory of program refinement

Summary: The refinement calculus is a logical framework for reasoning about programs. It is concerned with two main questions: is a program correct with respect to a given specification, and how can we improve, or refine, a program while preserving its correctness. Both programs and specifications can be seen as special cases of a more general notice that of a contract between independent agents. Refinement is defined as a relation between contracts and turns out to be a lattice ordering. The refinement calculus is formalized within higher-order logic. This allows us to prove the correctness of programs and to calculate program refinements in a rigorous, mathematically precise continent. Lattice theory and higher-order logic together form the mathematical basis for the calculus.

Literature:

1. Ralph-Johan Back, Joakim von Wright. Refinement Calculus: A Systematic Introduction. Springer-Verlag, 1998. DOI: 10.1007/978-1-4612-1674-2

2. Viorel Preoteasa and Ralph-Johan Back. Data Refinement of Invariant Based Programs. In Electronic Notes in Theoretical Computer Science, pages 143-163, Elsevier, 2009. DOI: 10.1016/j.entcs.2009.12.022

3. Linas Laibinis. Mechanised Formal Reasoning About Modular Programs. PhD thesis, TUCS Dissertations, Turku Centre for Computer Science, April 2000.

4. Anton Tarasyuk, Elena Troubitsyna, and Linas Laibinis. Integrating Stochastic Reasoning into Event-B Formal Development. Formal Aspects of Computing, 27(1), pages 53-77, 2015.

October 21, 2020. Speaker: doctoral students of the institute.

2020 09 23. Pranešėjas: dr. Linas Petkevičius

Report topic: Outliers identification in data

Summary: The data analysis and automated solutions are surrounding us.
However, usually model parameters estimation assumes that data are “clean”.
It might lead to biased estimates of parameters and misleading conclusions.
In seminar the formulation of outliers for scale-location distributions family are presented [1],
as well as for regression tasks [2].

Literature:

[1] V. Bagdonavičius, L. Petkevičius. “Multiple outlier detection tests for parametric models.” arXiv preprint arXiv:1910.10426 (2019).
[2] V. Bagdonavičius, L. Petkevičius. “A new multiple outliers identification method in linear regression.” Metrika 83.3 (2020): 275-296.