Lunch Seminar

The purpose of the UEF data-driven lunch seminar series is showcase state-of-the-art empirical research in the university. They also foster interdisciplinary networks around such empirical research and strengthen the visibility of the UEF research activities nationally and internationally.

The seminars typically take place on the last Friday of each month (12noon–1pm). They involve one invited and high-profile guest speaker per session, and the purpose is to invite guest speakers whose topics cross disciplinary boundaries and who offer food for thought for both specialists and generalists in data-driven research. The objective is that the lectures are informative and yet informal with a focus on outreach across the four faculties. Each session includes time reserved for questions and discussion.

The lectures take place in an online environment. The lectures are free and open for the UEF community. Each session has a named chairperson, who introduces the speaker and moderates the discussion.

Links to the talks are distributed in UEF communication channels (Heimo event calendar and Viva Engage). Members of UEF can also join Data-Driven lunch email group in Outlook to receive notifications for upcoming events. If you do not have access to the event link but would like to attend, please contact the session’s chairperson.

Spring 2024

26 April: Drug-induced transcriptomics for drug discovery using both AI and classical approaches

Assistant Prof. Francesco Napolitano, University of Sannio, Benevento, Italy

In recent decades, computational methodologies have emerged as pivotal instruments within the domain of drug discovery, especially though the so-called ligand-based and rational approaches. Nonetheless, there have been remarkable efforts towards the creation of comprehensive databases collecting drug-induced transcriptomic alterations. These repositories have laid the groundwork for the emergence of innovative, unbiased methodologies, which facilitate novel explorative tools in domains where traditional modalities have been already extensively attempted. Furthermore, the advent and widespread adoption of single-cell sequencing technologies have led to the generation of datasets of unprecedented scale. These extensive datasets serve as a fertile ground for the application of data-intensive artificial intelligence models, thereby heralding the incorporation of advanced machine learning techniques into the realm of drug discovery. This presentation aims to elucidate such developments, highlighting their conceptual underpinnings and showcasing significant applications that exemplify their potential.

Speaker of the Day

Francesco Napolitano

Assistant Professor

University of Sannio, Benevento, Italy

F.N. is currently an Assistant Professor of Bioinformatics at the University of Sannio, Benevento, IT, after major experiences at the Telethon Institute of Genetics and Medicine (TIGEM), Naples, IT, and the King Abdullah University of Science and Technology (KAUST), SA. His main research interests concern -omics data analysis for drug discovery, including small molecules for laboratory applications, drug repurposing, and elucidation of drug mode of action. In particular, he has published relevant literature in the area of machine learning techniques applied to the identification of small molecules inducing cell reprogramming, differentiation or rejuvenation from exhaustion, using integrated multi-omics data and deep learning models for domain adaptation. His additional interests include the application of omics data analysis to the study of genetic diseases, cancer, and metabolism through flux balance analysis. F.N. developed and published software tools in the areas of bioinformatics, data management and reproducible research, including R/Bioconductor packages. His main research projects have been funded by Fondazione Veronesi, KAUST, and the Italian Ministry for Universities and Research.

22 March: Data-driven approaches to multimodal communication

Prof. Tuomo Hiippala, University of Helsinki, Finland

Human communication naturally combines multiple ‘modes’ of expression, as exemplified by coordinated use of spoken language, gestures, posture and gaze in face-to-face interaction and the co-deployment of written language, diagrams, photographs, etc. in everyday media. This phenomenon, known as multimodality, is currently receiving increased interest in diverse fields of study ranging from linguistics to artificial intelligence research. In this presentation, I discuss the challenges of pursuing data-driven research on multimodality, especially from the perspective of multimodal semiotics, which studies how humans make and exchange meanings multimodally.

Chair: Paula Rautionaho

Speaker of the Day

Tuomo Hiippala

Associate Professor

University of Helsinki

Tuomo Hiippala is Associate Professor in English Language and Digital Humanities in the Department of Modern Languages at the University of Helsinki, where he leads the Multimodality Research Group. His current research interests include multimodal corpora and the use of computational methods in multimodality research. In 2023, he received an ERC Consolidator Grant for a project that develops novel methods for the empirical study of multimodality.

23 February: Brain MRI Analysis

Prof. Jose V. Manjón, Polytechnic University of Valencia, Spain

Quantitative MRI analysis is becoming fundamental for improving health care by providing earlier biomarkers to help in the decision making process. Advances in medical image processing and analysis are fundamental to support this area. This lecture presents some of the most important problems in brain MRI analysis and will describe one of the first online quantitative brain analysis softwares, volBrain https://www.volbrain.net/.

Chair: Jussi Tohka

Speaker of the Day

Jose V Manjón

Professor

Polytechnic University of Valencia, Spain

Professor Manjón is a professor of the Polytechnic University of Valencia. He is doctor in artificial intelligence and pattern recognition and the head of MIALAB–ITACA group during the last 20 years. His primary research interest include medical image processing with special emphasis on brain MRI. He has proposed state-of-the-art methods in different topics such as image denoising, superresolution, segmentation or classification. He is one of the creators of the volBrain platform that has analyzed online more than 500.000 brains worldwide. Recently, he has been selected as top 1% of world researchers by Clarivate 2022.

Autumn 2023

24 November: Big Data in Travel Mobility

Prof. Sangwon Park, Kyung See University, South Korea

The advancement of information technology allows tourism researchers to collect big data, which provides an innovative approach to exploring tourism. This lecture presents an understanding of mobile and/or GPS data as a kind of big data source and how the location data can be analyzed to generate important insights of tourist behaviors.

Speaker of the Day

Sangwon Park

Professor

Kyung See University

Professor Sangwon Park is a Professor at the College of Hotel and Tourism Management at Kyung Hee University. Professor Park has taught at the University of Surrey in the UK, and has served as Deputy Head of Hospitality Management, and the Hong Kong Polytechnic University. Professor Park has been awarded a PhD Degree from Temple University, USA (Business Administration) and received Master Degree in Hotel & Restaurant Management with a Statistics minor from the University of Missouri – Columbia, USA. His primary research interests include information technology in tourism and hospitality, tourism big data/AI, and digital marketing. He received an award of Emerging Scholar of Distinction at International Academy of the Study of Tourism in 2019 and recently, has been selected as top 1% of world researchers in social science by Clarivate 2022.

27 October: GEDI: Data-Driven Developments in Forest Biomass Surveys

Dr. Svetlana Saarela, Norwegian University of Life Sciences, Ås, Norway

The Global Ecosystem Dynamics Investigation (GEDI), a high-resolution spaceborne laser developed by NASA, was operational on the International Space Station (ISS) from December 2018 to January 2023. During these years, GEDI generated over 15 billion cloud-free observations across Earth’s tropical and temperate regions. While the sensor is currently dormant, it is expected to be brought back to use early next year.

The GEDI mission offers a range of data products, including single footprint and grid data, which describe Earth’s 3D features. One of these products, known as the Gridded Above Ground Biomass Density (AGBD) product or the L4B product, aims to develop a global biomass map with a spatial resolution of 1 km. The mission employs hybrid and hierarchical model-based (HMB) statistical inference methods. These methods allow for the assessment of forest biomass and associated uncertainties using GEDI data alone (Hybrid inference) or in combination with data from NASA’s Landsat mission (HMB inference).

I will be presenting novel developments in the hybrid and HMB inferential methods, which incorporate nonparametric machine learning techniques. The primary focus of the presentation will be on uncertainty assessment using a bootstrapping procedure that separates the computations into parts, thus reducing the computational time required. This approach makes bootstrapping more attractive to users.

Speaker of the Day

Svetlana Saarela

Norwegian University of Life Sciences

Dr. Svetlana Saarela defended her doctoral thesis at the University of Helsinki in 2015. Currently, she is employed as a researcher at the Norwegian University of Life Sciences in Ås, Norway. Her main research interest is developing statistical frameworks for forest inventories by combining remotely sensed data from different sensors with field sample data. Since 2018, her primary work for the GEDI mission has been to adapt the hybrid and HMB statistical inference methodologies for the mission’s biomass map product (L4B).

Spring 2023

28 April: Human-Centered AI to foster Explainability and Robustness for ensuring Trustworthy AI

Professor Andreas Holzinger, Human-Centered AI Lab, University of Natural Resources and Life Sciences Vienna, Austria

Recent advancements in statistical machine learning have resulted in the resurgence of interest in artificial intelligence (AI). However, there are two key areas that require significant improvement, namely robustness and explainability. In many application domains, the reason behind the obtained result is often more crucial than the result itself. This is closely linked to robustness, as even slight changes in the input data can lead to substantial impacts on the output and entirely different results. This is of utmost importance in all critical areas where we use real-world data from our surroundings, as opposed to laboratory data that is independent and identically distributed (i.i.d). As a result, the use of AI in domains that affect human life, such as agriculture, climate, forestry, and health, has resulted in an increased need for trustworthy AI. In sensitive areas where accountability, transparency, and interpretability are required, explainable AI (XAI) is now essential due to regulatory obligations. One approach to making AI more robust is to merge statistical learning with knowledge representations. In certain tasks, it can be advantageous to involve a human in the loop. A human expert can sometimes provide valuable experience and conceptual understanding to the AI pipeline.

Andreas Holzinger pioneered in interactive machine learning with the human-in-the-loop promoting robustness and explainability to foster trustworthy AI. He advocates a synergistic approach of Human-Centered AI (HCAI) to put the human in-control of AI, aligning artificial intelligence with human intelligence, human values, ethical principles, and legal requirements to ensure secure and safe human-machine interaction. For his achievements he was elected a member of Academia Europaea in 2019, the European Academy of Science, of the European Laboratory for Learning and Intelligent Systems (ELLIS) in 2020, and Fellow of the international federation of information processing (ifip) in 2021. Andreas Holzinger serves as consultant for the Canadian, US, UK, Swiss, French, Italian and Dutch governments, for the German Excellence Initiative, and as national expert in the European Commission (EC). Andreas is in the advisory board of the Artificial Intelligence Strategy AI made in Germany 2030 of the German Federal Government. He obtained his Ph.D. with a topic in Cognitive Science from Graz University in 1998, and his Habilitation (UOG 93, venia docendi) in Computer Science from Graz University of Technology in 2003. Andreas was Visiting Professor for Machine Learning & Knowledge Extraction in Verona (Italy), RWTH Aachen (Germany), and the University College London (UK). From July 2019 until February 2022 Andreas was a Visiting Professor for explainable AI at the University of Alberta (Canada). Andreas Holzinger has been appointed full professor for digital transformation in smart farm and forest operations at the University of Natural Resources and Life Sciences Vienna and had his inaugural lecture on November 7, 2022 – see: https://human-centered.ai/antrittsvorlesung-andreas-holzinger

31 March: Microsimulation as a tool for policy evaluation

Research Manager Jussi Tervola, Finnish Institute for Health and Welfare

Income transfers are a focal feature of contemporary societies and policies. Social security and taxation systems form a complex net both at individual and societal level. Different policies interact in a way that is impossible to grasp without advanced analysis tools. Microsimulation is a type of computer modeling technique used in social sciences and economics to simulate these interactions with micro-level population data. Microsimulation is used widely in ministries and governmental research institutes to evaluate different policy alternatives as well as past reforms. In academia, microsimulation is used above all in comparative policy analysis. Microsimulation can be used, for example, to evaluate the future and past policy effects on income inequality, national budgets, or employment. In this talk, I give an overview of the possibilities and limits of microsimulation tools in policy analysis.

Bio: Jussi Tervola (Dr.Soc.Sci) is currently a research manager at the Finnish Institute for Health and Welfare (THL), an expert and research institute that provides information for decision-making in the field of health and welfare. He specializes in quantitative and comparative policy analysis as well as the microsimulation method. He has been the chair in multiple expert groups that analyze the future scenarios or past reforms of social security in Finland. Prior to THL, he has worked as a guest researcher in Stockholm University and as a researcher in the Social Insurance Institution of Finland (Kela). He received his doctoral degree in 2018 in the field of social and public policy from the University of Helsinki.

24 February: Linguistic perspectives to politics and culture

Prof. Jukka Tyrkkö, Linnaeus University

Language plays a central role in nearly all human endeavours, and consequently the textual and linguistic analysis of the written and spoken record of our common heritage can play a key role in uncovering broader diachronic and synchronic tendencies and trends that may otherwise escape our attention. In this talk, I will focus on the language of politics and the many ways in which the systematic study of political language over time and contrastively across contexts can reveal how both elected and unelected leaders manufacture consent and manipulate public opinions. Over the course of the presentation, I will refer to both recently published studies and present some new findings, as well as discuss the general principles of studying political language use and cultural phenomena with the aid of linguistic corpora. Some of the topics covered will include changes in the use of inclusive and exclusive language use, the ways in which major crises and developments in media affect political speaking, the uses of repetition and other rhetorical tropes, and how the language of civil rights activists differs from elected politicians

Bio: Jukka Tyrkkö is Professor of English Linguistics at Linnaeus University. He has previously served as Professor of English at the University of Tampere and a Visiting Professor in Digital Humanities at the University of Turku, and later in the spring of 2023 he will be Visiting Professor at Heidelberg University. Jukka has primarily worked in diachronic corpus linguistics, focusing on a variety of fields including medical writing, political discourse, lexicography, and computer-mediated communication. His work is primarily quantitative in nature, with a particular interest in lexical and phraseological patterns. He is currently an executive board member of ICAME, chair of the Helsinki Society for Historical Lexicography, and series editor of Language, Data Science and Digital Humanities (Bloomsbury Academic) with Prof. Mikko Laitinen.

27 January: Modern speech-recognition technology in action: a data-driven revolution

Dr. Antti Ukkonen, Speechly

Automatic speech-recognition (ASR) has historically had a reputation of “almost working, but not really.” However, during the past 10 years, ASR technology has taken immense leaps ahead. Home-automation systems and personal assistants such as Siri have reached a level of performance that is most of the time practically usable, rather than merely annoying. These developments have been driven less by advances in algorithms or theory, but more by availability of data and efficient infrastructure for computation. In this talk I give a brief general introduction to the topic of ASR, and outline some relevant research activities. I also discuss the ups and downs of life in a technology start-up that operates in an area where the next revolution is always right around the corner.

Bio: Antti is currently the Chief Product Officer at Speechly, a Helsinki-based start-up company focusing on speech-recognition technology. He spends his days on a wide spectrum of activities ranging from prioritising engineering activities and dealing with customers to writing C++ code for efficient machine learning inference on mobile devices. Prior to joining Speechly in 2020, Antti spent 15 years in academia doing research on data mining algorithms. After obtaining his doctoral degree at Aalto university in 2008, Antti has worked at Yahoo! Research, Helsinki Institute for Information Technology HIIT, Finnish Institute for Occupational Health, and finally as an Academy Research Fellow at University of Helsinki.

Autumn 2022

2 December: Big Data will Require Big Knowledge to be Useful for Future Personalized Medicine – What are the Computational Tools?

Dr. Sui Huang, Institute for Systems Biology, Seattle, WA, USA

Bio: Dr. Sui Huang is a molecular and cell biologist with a strong background in theoretical biology. He has devoted his research to understanding the very phenomenon of cancer from a complex systems perspective. Life scientists now readily acknowledge that the “whole is more than the sum of its parts” but the question is: What exactly is the “more” that we need in order to understand the “whole”? Can this abstract philosophical notion be reduced to a rigorous formal concept and concrete molecular entities? Pursuing this question has guided Dr. Huang‘s research in cancer and cell biology over the past decade. Before joining the ISB in fall 2011, Dr. Huang held faculty positions at the University of Calgary (Institute of Biocomplexity and Informatics), where he helped establish biocomplexity as a discipline in research and teaching, and at Harvard Medical School (Children’s Hospital) where he obtained first experimental evidence for the existence of high-dimensional attractors in mammalian gene regulatory networks.

Sui Huang grew up in Geneva and Zurich. He received his MD degree from the University of Zurich and obtained thereafter, as the first recipient of the PhD-Program-for-Physicians Award of the Swiss National Science Foundation, his PhD in molecular biology and physical chemistry for work on interferons. As a postdoctoral fellow at Children’s Hospital Boston he investigated tumor angiogenesis and cell growth control. In that period he also studied dynamical systems through his affiliation with the New England Complex Systems Institute.

Seeing how both interferons and anti-angiogenic agents have failed to live up to their celebrated promise of curing cancer has had a lasting impact on Dr. Huang’s views. The humbling recognition of the profound complexity of the living state fostered the desire to overcome the orthodoxy of reductionist, monocausal and deterministic thinking that prevailed in biomedicine and to put to use his knowledge of complex systems theory in his experimental research. Time was ripe in the late ’90s because the arrival of the “omics technologies” and systems biology paved the way towards this interdisciplinary approach. With his move to the ISB, Dr. Huang continues to unite experiment and theory to gain insights in the essence of multi-cellularity and cancer.

30 September: The Lecture of Why

Professor Jilles Vreeken, CISPA Helmholtz Center for Information Security, Saarbrücken, Germany

Determining cause from effect is perhaps the most fundamental problem in science. So far, we have always had to ponder deeply to come up with good hypotheses, and then set up expensive experiments to verify those. Wouldn’t it be great if we could just use machine learning to extract true causal dependencies from data that we gathered outside of a controlled experiment?

In this talk, I will first disappoint you, as I have to start with the cold hard truth that data alone is insufficient to draw causal conclusions. Simply put, with observational data alone we will never be able to tell the true relationship between murder rates and ice cream sales. I will then make amends, and explain how we can discover causal dependencies if we make assumptions about how we think the world works.

In particular, I will explain how we can discover causal graphs using the idea of explaining-away, and how we can determine causal directions using Occam’s razor. If time permits, I will also answer the age-old question of whether per-capita chocolate consumption causes Nobel prizes or not.

Bio: Jilles Vreeken is tenured faculty at the CISPA Helmholtz Center on Information Security, where he leads the Exploratory Data Analysis group. He is an Honorary Professor at Saarland University and a Senior Researcher at the Max Planck Institute for Informatics.

His research interests include data mining, machine learning, and causal inference. He is particularly interested in developing well-founded theory and efficient methods for extracting causal models and informative patterns from complex data, as well as in how to put these to good use. He has authored 3 book chapters and over 110 conference and journal papers. He received three best paper awards, the ACM SIGKDD 2010 Doctoral Dissertation Runner-Up Award, and the IEEE ICDM 2018 Tao Li Award.

Spring 2022

29 April: The European Social Survey: Monitoring Social Change in Europe since 2002

Professor Heikki Ervasti, University of Turku

The European Social Survey (ESS) is an academically driven comparative social survey designed to chart and explain the interaction between Europe’s changing societies and the attitudes, beliefs and behaviour patterns of its diverse populations. Since 2013, the ESS has held the European Research Infrastructure Consortium (ERIC) status.

The ESS forms a biennial time series starting from 2002, now working on the 11th Round in 2022/23. Within the past two decades, the ESS has acquired an established position as one of the most important data sources in social sciences and related fields. A multidisciplinary research community consisting of approximately 200.000 social and political scientists, statisticians, economists, geographers, psychologists, health researchers, etc., from all over the world use the ESS data. The ESS data covers more than 35 European and surrounding nations.

The ESS employs extremely rigorous methodological standards in sampling, question-testing, translation and field-work procedures, and continuously develops survey methods and conducts methodological research. The data sets are immediately and freely available in the internet for all with no restrictions. The ESS has served as a data source for thousands of journal articles, conference papers, books and other publications.

In my presentation, I will review the starting points of the ESS, methodological features of the ESS, discuss the multiple ways of using the data, and show some examples of recent ESS-based research, especially in the field of welfare state research.

Bio: Heikki Ervasti is professor of social policy in the University of Turku. He has worked within the ESS as the Finnish PI and National Coordinator and since 2002, and the Finnish PI in CRONOS-2, which is the first comparative internet panel survey based on representative random samples in the 9 participating countries. Ervasti’s main research interests include Europeans’ social and political attitudes, wellbeing in comparative perspective, and comparative welfare state and labour market research.

25 March: Computational approaches for single-cell research

Professor Laura Elo, University of Turku

We develop computational data analysis tools and mathematical modelling methods for analyzing and interpreting data generated by modern high-throughput biotechnologies, such as next-generation sequencing and mass spectrometry-based proteomics. A specific focus is on biomedical applications in close collaboration with experimental and clinical groups to enable robust and reproducible interpretation of the molecular as well as clinical data. Using statistical modelling and advanced machine learning techniques, we have, for instance, identified early markers for type 1 diabetes and developed several powerful computational models for predicting disease and treatment risks. Our ultimate goal is to improve the diagnosis, prognosis and treatment of complex diseases, such as diabetes and cancer.

Bio: Laura Elo is Professor of Computational Medicine and Head of Medical Bioinformatics Centre at University of Turku, Finland. She has PhD in applied mathematics and long experience in molecular systems immunology and application of machine learning in medicine. Her research group develops computational methods to interpret molecular and clinical data in several research projects (e.g. ERC). In 2019 she received the L’Oréal-UNESCO International Rising Talent Award. She has published >130 research articles and several software packages and is member of various boards.

25 February: Supporting ethnographic social media research with computational analytics: case Christchurch

Senior Research Fellow Jukka Huhtamäki, Tampere University

Qualitative and quantitative research methods are often considered not only as alternatives but different ends of a continuum. This presentation paints a different picture in showing how computational methods can be used to support ethnographic research. Specifically, the presentation gives a hands-on overview of four years of research on violent hybrid media events that combine ethnographic media research with computational methods, including collecting data from Twitter, using topic modeling as means for distant reading of the data, and social network analysis for insights on the structure of the interactions in the data. Examples taking a literate programming approach in the form of Jupyter Notebooks are given to illustrate the day-to-day interdisciplinary working practices when studying the aftermath of Christchurch mosque attacks in the hybrid media system.

Bio: Jukka Huhtamäki (D.Sc. (Tech.)) is a senior research fellow at the unit of knowledge and information management at Tampere University. He specializes in social network analysis, visual analytics, and computational methods. Recently, he has developed and applied these methods in studying fluid organizing and the hybrid media system. @jnkka is on Twitter.

28 January: The benefit of subject-matter theory in model building: the case of tree stem volume

Prof. Lauri Mehtätalo, Natural Resources Institute Finland (Luke), Joensuu

Models for the volumes of tree stems are one of the most basic models needed in practical forestry and are needed, for example, in estimating the total volume of growing forest trees before harvest. Tree volume cannot be easily measured. Therefore, volume functions that express the stem volume based on tree diameter at breast height (DBH) and total height are needed. In Finland, the currently applied volume functions were published already 40 years ago. Since that, the form of tree stems has changed because of forest management practices and tree breeding, for example and new models are needed. Tree stems are three-dimensional objects, and their shape is somewhere between a cone and a cylinder. Based on this, I develop a theoretically justified volume function and fit it to a data set. The data includes the trees used in the old volume models (collected by climbing to the trees in 1970’s), a data collected in 1990 by felling the trees, and a recent data set based on terrestrial laser scanning of standing trees. The data includes a total of 8500 trees. Comparison to an data-driven model formulation based on empirically found functions is also presented.

Bio: Lauri Mehtätalo works currently as research professor in mathematical modelling for forest planning at Natural Resources Institute Finland (Luke). Before moving to Luke, he worked as professor in applied statistics at UEF. He did his PhD in forest mensuration and has also worked as senior researcher in forest planning at UEF. He is the main author of a recent textbook monograph Mehtätalo and Lappi 2020. Biometry for Forestry and Environmental Data: With Examples in R.

Autumn 2021

3 December: Spatiotemporal Machine Learning for Football

Prof. Ulf Brefeld, Leuphana Universität Lüneburg, Germany

Data analyis in team sports is a complex task and involves observations of agents (i.e., the players) and their actions (passing, running, etc.) as well as additional variables (score, ball possession, etc.) over time (the duration of the game) and space (the pitch). On the example of football, I will introduce how to make sense of those observations and in particular introduce movement models as key to every higher level analysis. I will also introduce some exemplary results that make use of movement models to bridge the gap towards addressing semantics of the game (e.g., next action, availability for pass).

Bio: Ulf Brefeld is a professor for Machine Learning at Leuphana Universität Lüneburg. Prior to joining Leuphana, he was a joint professor for Knowledge Mining & Assessment at TU Darmstadt and the German Institute for Educational Research (DIPF), Frankfurt am Main. Before, he led the Recommender Systems group at Zalando SE and worked on machine learning at Universität Bonn, Yahoo! Research Barcelona, Technische Universität Berlin, Max Planck Institute for Computer Science in Saarbrücken, and at Humboldt-Universität zu Berlin. Ulf received a Diploma in Computer Science in 2003 from Technische Universität Berlin and a Ph.D. (Dr. rer. nat.) in 2008 from Humboldt-Universität zu Berlin. He is interested in statistical machine learning and data mining.

29 October: How Old is Your Brain? BrainAGE as Neuroimaging Biomarker

Prof. Christian Gaser, Jena University Hospital, Germany

In this talk I will briefly introduce a method that allows to estimate the deviation from normal brain maturation and aging on an individual level. We use a T1-weighted magnetic resonance (MR) images to predict the individual brain age. The difference between the estimated and the chronological age is termed as brain age gap estimation (BrainAGE) score and indicates the degree of abnormal brain aging. Consequently, the BrainAGE score directly quantifies the amount of deviation from normal brain aging (in years). For example, if a 70-year-old individual has a BrainAGE score of +5 years, this means that this individual shows the typical atrophy pattern of a 75-year-old individual.
I will give a short overview about the underlying method and will finally demonstrate some applications of this approach.

Bio: Christian Gaser is an Associate Professor of Computational Neuroscience/Neuroimaging at the Jena University Hospital in Germany. He studied Electrical Engineering and Technical Acoustics in Chemnitz and Dresden and received his PhD degree in Neuroscience from the University of Magdeburg, Germany, in 2001. He is currently head of the Structural Brain Mapping Group at the Jena University Hospital. His research program is focused on the development of advanced computational tools for the analysis of structural brain data. In particular, he develops and applies algorithms and tools for processing voxel- and surface-based imaging data that include segmentation, surface reconstruction, and disease prediction. Moreover, he has developed a framework that automatically and reliably provides brain age estimates (BrainAGE) for individuals based on their MR-images. He has also developed several software tools, including the Computational Anatomy Toolbox for performing voxel- and surface-based morphometry, which is widely used in the scientific community.

1 October: The Beauty of Studying Fertility and Family Dynamics with Population Register Data

Dr. Marika Jalovaara, University of Turku, Finland

For decades, fertility levels in the Nordic countries have been high by European standards. This has been attributed to the Nordic welfare states’ support for families and gender equality. The unprecedented Nordic fertility decline since 2010 confronts the scientific paradigm of what drives fertility, and creates major policy challenges for already rapidly ageing societies. Further, family and fertility dynamics are linked to social and gender inequalities, and these links – often reciprocal – contribute to the accumulation of disadvantages across life courses and generations.

Our current research aims to increase understanding of changing fertility and family dynamics and their interlinkages with social and gender inequalities in modern welfare societies, as manifested by the case of Finland and the other Nordic countries.As a warmup, we take a look at fertility trends in Finland. Then, this presentation introduces some of our recent and ongoing research on fertility and family dynamics in Finland. We use population register data, and go beyond traditional fertility research by including gender comparisons, and data on coresidential partnerships and couples. In addition to standard demographic (e.g., event history methods), we use sequence methods, which allow a more holistic and exploratory approach to fertility and family dynamics.

Bio: Marika Jalovaara is a demographer and sociologist, whose main interests include links among family dynamics (cohabitation, marriage, childbearing) and social inequalities. She received her PhD in 2007 at the University of Helsinki, and since 2012, she is an associate professor (docent) in demography at the University of Helsinki and in economic sociology at the University of Turku. She is affiliated at the University of Turku, and is currently the PI of the project ‘Falling Fertility and the Inequalities Involved, NEFER‘, 2019–2023, funded by the Academy of Finland, and Co-PI, Research Area Director (Area: Demography and Life Course) and Senior Research Fellow in INVEST Research Flagship ‘Inequalities, Interventions, and the New Welfare State’. Starting Oct 2021, she is the PI and director of the consortium “Family Formation in Flux – Causes, Consequences and Possible Futures (FLUX)”, funded by the Strategic Research Council, Academy of Finland. She also serves as the President of the Finnish Demographic Society.

Spring 2021

28 May: Visualization of Multidimensional Data: Possibilities, Pitfalls, and Practices

Prof. Andreas Kerren, Linköping University and Linnaeus University, Sweden

This seminar will introduce the main ideas of interactive data visualization with a special focus on multidimensional data and applications. The field of Information Visualization (InfoVis) centers on abstract data without spatial correspondences; that is, usually it is not possible to map this information to the physical world. Examples of such abstract data are tabular, networked, hierarchical, or textual information sources. The related field of Visual Analytics (VA) focuses on the analytical reasoning of typically large and complex (often heterogeneous) data sets and combines techniques from interactive visualizations with computational analysis methods. I will show how these two fields belong together, their potential to efficiently analyze large and complex multidimensional data sets, and pitfalls that should be avoided when applying techniques from these fields. Finally, I will showcase related InfoVis and VA approaches in more detail where multidimensional data is analyzed together with other data types, such as text or networks.

Bio: Dr. Andreas Kerren received his PhD degree in Computer Science from Saarland University, Saarbrücken, Germany. In 2008, he achieved his habilitation (docent competence) from Växjö University, Sweden. Dr. Kerren is currently a Full Professor of Information Visualization, Linköping University (LiU) and Linnaeus University (LNU), Sweden. He holds the Chair of Information Visualization at LiU and is head of the research group Information and Software Visualization at LNU. In addition, he is an ELLIIT professor supported by the Excellence Center at Linköping–Lund in Information Technology and key researcher of the Linnaeus University Centre for Data Intensive Sciences and Applications. His main research interests include several areas of information visualization and visual analytics, especially visual network analytics, text visualization, and the use of visual analytics for explainable AI. He is editorial board member of the Information Visualization and Computer Graphics Forum journals, has served as organizer/program chair at numerous conferences, such as IEEE VISSOFT 2013/2018 or GD 2018, and has edited a number of successful books on human-centered visualization. Kerren has published about 200 peer-reviewed papers, articles, and book chapters.

23 April: Emotion AI: Research and Applications

Prof. Guoying Zhao, University of Oulu

Emotions play a key role in human-human interactions and become one key focus in future Artificial Intelligence. There is a growing need to develop emotionally intelligent interfaces, which are able to read the emotions of the users and adapt their operations accordingly. Among the areas of application are human-robot interaction, emotional chatpots, health and medicine, on-line learning, user or customer analysis, and security and safety. Face analysis based on computer vision will play a key role in developing such interfaces. This talk will provide an introduction to the emotional interfaces, and overviews our recent progress in related research. The research topics to be covered include facial expression recognition, analysis of micro-expressions, emotional gesture study, remote heart rate measurement from videos and potential applications. Finally, some future challenges are outlined.

Bio: Guoying Zhao received the Ph.D. degree in computer science from the Chinese Academy of Sciences, Beijing, China, in 2005. Then she worked as senior researcher since 2005 and an Associate Professor since 2014 with the Center for Machine Vision and Signal Analysis, University of Oulu, Finland. She is currently a full professor with University of Oulu, Finland from 2017, and a visiting professor with Northwest University, China from 2016. In 2020, she was selected for the prestigious post of Academy Professor with the Academy of Finland 2021-2026. She is IAPR Fellow since 2020, and was Nokia visiting professor in 2016. She has authored or co-authored more than 255 papers in journals and conferences. Her papers have currently over 15000 citations in Google Scholar (h-index 56). She is co-program chair for ACM International Conference on Multimodal Interaction (ICMI 2021), was co-publicity chair for FG2018, General chair of 3rd International Conference on Biometric Engineering and Applications (ICBEA 2019), and Late Breaking Results Co-Chairs of ICMI 2019, has served as area chairs for several conferences and is associate editor for Pattern Recognition, IEEE Transactions on Circuits and Systems for Video Technology, and Image and Vision Computing Journals. She has lectured tutorials at ICPR 2006, ICCV 2009, SCIA 2013 and FG 2018, authored/edited three books and nine special issues in journals. Dr. Zhao was a Co-Chair of many International Workshops at ICCV, CVPR, ECCV, ACCV and BMVC. Her students and researchers are frequent recipients of very prestigious and highly competitive fellowships, such as Academy of Finland Postdoc position, the Nokia Scholarship, Endeavour Research Fellowship, Tauno Tönning Research funding, Kauta Foundation grant and Jorma Ollila grant. Her current research interests include image and video descriptors, facial-expression and micro-expression recognition, emotional gesture analysis, affective computing, and biometrics. Her research has been reported by Finnish TV programs, newspapers and MIT Technology Review.

26 March: Computational Problems in Mining Social Media

Prof. Aristides Gionis, KTH Royal Institute of Technology, Sweden

Online social media is an important venue of public discourse today, hosting the opinions of hundreds of millions of individuals. Social media are often credited for providing a technological means to break information barriers and promote diversity and democracy. In practice, however, the opposite effect is often observed: users tend to favor content that agrees with their existing world-view, get less exposure to conflicting viewpoints, and eventually create “echo chambers” and increased polarization. Arguably, without any kind of moderation, current social-media platforms gravitate towards a state in which net-citizens are constantly reinforcing their existing opinions.

In this talk we present some of our ongoing work in social media mining. We first focus on the problem of detecting polarization in signed networks, which offer a simple but powerful abstraction to model user interactions by annotating edges as positive (friendly) or negative (antagonistic). Detecting polarization in signed networks is formulated as searching for two subsets of vertices (communities) having mostly positive edges within and mostly negative edges across. We distinguish different problem variants, and we develop algorithms with provable guarantees based on spectral analysis. We then address the problem of designing algorithms to break filter bubbles, reduce polarization, and increase diversity. We discuss different strategies based on content recommendation and increasing diversity.

Bio: Aristides Gionis is a WASP professor in KTH Royal Institute of Technology and adjunct professor in Aalto University. He is currently a fellow in the ISI Foundation, Turin, and in 2016 he was a visiting professor in the University of Rome. His previous appointment was with Yahoo! Research, Barcelona. He obtained his PhD in 2003 from Stanford University. He is currently serving as an action editor in the Data Management and Knowledge Discovery journal (DMKD) and an associate editor in the ACM Transactions on the Web (TWEB). He has contributed in different areas of data science, such as algorithmic data analysis, web mining, social-media analysis, data clustering, and privacy-preserving data mining.

5 March: Human Brain Mapping with Data-Driven Techniques

Prof. Lauri Nummenmaa, University of Turku