Dr. Hossein Bonakdari, Ph.D., P.Eng., esteemed professor at the University of Ottawa, Canada, stands at the forefront of developing artificial intelligence techniques in Time Series analysis. His innovative integration of stochastic and machine learning algorithms has opened new pathways in addressing time series problems, significantly impacting both national and international research landscapes. Dr. Bonakdari’s pioneering work focuses on crafting cutting-edge models for real-time data analysis and forecasting, contributing profoundly to the field. His seminal work, "Stochastic Modeling: A Thorough Guide to Evaluate, Pre-Process, Model, and Compare Time Series," paves the way for groundbreaking approaches in time series analysis and predictive modeling, particularly in applying stochastic models.
A prolific author, he has published over 280 papers in indexed journals, authored 14 book chapters, and penned three books. According to a Scopus report, his scholarly contributions are widely recognized and revered, amassing over 6500 citations with an impressive h-index of 46. An acclaimed speaker, Dr. Bonakdari has delivered over 100 presentations at various national and international conferences, sharing his insights and research advancements.
Dr. Bonakdari is a respected section Board Member of the Sustainability Journal, and serves on the Editorial Boards of the Natural Resources Journal and the Earth Journal, further contributing to the scholarly community. Dr. Bonakdari's exceptional research achievements have garnered global recognition, consistently placing him in the top 2% of the world's top scientists across various fields for four consecutive years (2019-2023).
Prof. Faramarz Famil Samavati, Professor, Department of Computer Science, University of Calgary .
Dr. Faramarz F. Samavati is a full professor of Computer Science at the University of Calgary. His primary research focus lies in Computer Graphics, Visualization, Digital Earth, and Scientific Computations. While achieving success in theoretical research, he actively pursues real-world and practical applications related to his fundamental research. His efforts on theory to real-world applications have made solid contributions to multiple technological innovations through various approaches, including industrial outreach, commercialization, tech transfer, and the development of software tools.
Dr. Samavati has been recognized with numerous awards, including the Peak Scholar Award for his excellence in Innovation and Knowledge Engagement, the Established Career Scholarship Excellence Award, the Digital Alberta Award, and the Association for Computing Machinery (ACM) Recognition of Service Award. Dr. Samavati leads the Graphics, Interaction, and Visualization (GIV) research lab and supervises a large team of graduate students and research staff. With over 150 peer-reviewed publications, Dr. Samavati has received eight Best Paper Awards for his innovative research.
Prof. Eamonn Keogh is a Distinguished Professor at the Department of Computer Science and Engineering in the University of California Riverside
He has invented many of the most commonly used primitives and representations for time series data mining, including SAX, PAA, Time Series Shapelets, Time Series Discords, Time Series Motifs, Time Series Chains and the Matrix Profile. The majority of papers on time series published in SIGKDD/ICDM/SIGMOD/VLDB etc. exploit at least one of his ideas or definitions. He has won at least one best paper award in all the major conferences in his area, including SIGKDD, ICDM,SIGMOD, SDM etc. His work is heavily cited; he has an H-index of 108 and a citation count of ~56,000.
His research is widely used in industry, and has been funded by Google, Microsoft, IBM, Sony, Oracle, Mitsubishi, Vodafone, NetAPP, Samsung, AspenTech, Applied Materials and Siemens
Title of the presentation:
The title of this talk may seem like clickbait, but it asks a serious question. It is clear that deep learning has greatly contributed to artificial intelligence in general. And there are hundreds of papers on deep learning for time series problems…In spite of this, I claim that the evidence that deep learning is useful for time series problems is close to zero. I will demonstrate this claim with comprehensive experiments and demonstrations. I will explain why flawed benchmark datasets, flawed metrics of success, and generally sloppy thinking has led the community astray. I will end the talk with advice about what the community can do to prevent this mismatch between our claims and reality.
Martin Wagner currently is Professor of Economics at the University of Klagenfurt, Chief Economic Advisor at the Bank of Slovenia and Fellow of the Macroeconomics and Economic Policy group at the Institute for Advanced Studies, Vienna. From October 2017 until end of 2018 he was Chief Economist of the Bank of Slovenia, being on leave from Technical University Dortmund, where we has been Professor of Econometrics and Statistics in the Faculty of Statistics of the Technical University Dortmund from 2012 until 2019. He was educated in Vienna, at the Technical University and the Institute for Advanced Studies, obtaining Diplomas in Mathematics (1995) and Economics (1998), as well as his Doctorate (2000). He obtained his Habilitation in Economics in 2007 at the University of Bern. Martin Wagner has worked at the Technical University of Vienna, the Institute for Advanced Studies in Vienna, the University of Bern and has been Professor of Econometrics and Empirical Economics at the University of Graz before his arrival in Dortmund. Visiting positions have brought him to Princeton University and the European University Institute in Florence.
His work has been published, amongst other outlets, in Journal of Econometrics, Econometric Theory, Journal of Applied Econometrics, Econometric Reviews, Econometrics, Oxford Bulletin of Economics and Statistics, Journal of Empirical Finance, Economics of Transition and Ecological Economics.
Title of the presentation:
Authors: Karsten Reichold, Martin Wagner, Milan Damjanovic, Marija Drenkovska
This paper presents evidence for sources and channels of nonlinearities and instabilities of the new Keynesian Phillips curve (NKPC) for the euro area and all but four member states over the last two decades prior to the COVID-19 crisis. The approach rests upon misspecification testing using auxiliary regressions based on the standard open-economy hybrid NKPC. Using a large number of specifications, this approach allows to systematically, i.e., based on a literature review, disentangle the evidence for nonlinearities and instabilities of the NKPC according to sources and channels. For the euro area and most considered member states, there is substantial evidence for nonlinearities and instabilities. The relatively most important channels of nonlinearities and instabilities are similar across countries, whereas the relatively most important sources differ across countries. The results strongly indicate the need for considering nonlinear NKPC relationships in empirical analyses and also point towards potentially useful nonlinear specifications.
Keywords: Euro area, instability, new Keynesian Phillips curve, nonlinearity, specification analysis
Daniel Peña was born in Madrid in 1948. He received his PhD in Industrial Engineering from the Polytechnic University of Madrid, Bachelor’s Degrees in Sociology and Statistics from the Complutense University of Madrid and Business Administration ITP from Harvard.
President of Carlos III University of Madrid in the period 2007- 2011 and re-elected in March 2011. He was Director of the Management Committee (1993-2000) and Vice Rector of the Carlos III University of Madrid (1992- 1995), where he is Full Professor of the Department of Statistics. He has also been Full Professor at the Polytechnic University of Madrid, the University of Wisconsin- Madison and the University of Chicago. He is Founding Director of the Quantitative Methods Department of the EOI Business School, of the Statistics Laboratory of the ETSII-UPM (the Higher Technical School for Industrial Engineering at the Polytechnic University of Madrid), as well as the Department of Economics and Statistics and Econometrics of the Carlos III University of Madrid.
He has been Director of the “Revista Estadística Española” and President of the Spanish Society of Statistics and Operative Research (Sociedad Española de Estadística e Investigación Operativa). In the same field, he stands out as the Founding President of the Statistical Methods Committee of the Spanish Association for Quality and member of the State High Council of Statistics, Vice President of the Inter-American Institute of Statistics and President of European Courses in Advanced Statistics.
He has published thirteen books and more than 190 research articles on Statistics, Quality and its Applications. He is Associate Editor of several international journals and has received national and international awards for research. In 2006 he received the Youden Award for the best article published in “Technometrics”. He is an honorary member of prestigious international associations such as the Institute of Mathematical Statistics and the American Statistical Association.
Title of the presentation:
Clustering scalar time series can be carried out using their univariate properties and hierarchical methods, especially when the dynamic structure of the series is of interest. Two major issues in clustering analysis are to detect the existence of multiple clusters and to determine their number, if exist. In this paper we propose a new test statistic for detecting the existence of multiple clusters in a time-series data set and a new procedure to determine the number when clusters exist. The proposed method is based on the jumps, i.e., the increments, in the heights of the dendrogram when a hierarchical clustering is applied to the data. We use parametric bootstraps to obtain a reference distribution of the test statistics and propose an iterative procedure to find the number of clusters. The clusters found are internally homogeneous according to the test statistics used in the analysis. The performance of the proposed procedure in finite samples is investigated by Monte Carlo simulations and illustrated by some empirical examples. Comparisons with some existing methods for selecting the number of clusters are also investigated and extensions of the proposed method to clustering for dependency will be presented.
Keywords: Dendrogram, Distance, Gap statistic, Hierarchical clustering, Jump, Parametric bootstrap, Silhouette statistic, Similarity