Program

Contributed Talks

• Almeida, Dora (CEFAGE - Universidade de Évora); Dionisio, Andreia (CEFAGE); Vieira, Isabel (CEFAGE); Ferreira, Paulo (VALORIZA)
  Does the cryptocurrency markets are weak-form market inefficient? Insights from information, chaos theory, and detrended fluctuation analysis
  This paper tests the weak-form efficiency of the top 16 cryptocurrencies by market capitalization. We utilized mutual information, chaos theory, and detrended fluctuation analysis as the techniques used are adaptable to explain the complex cryptocurrency's behavior and efficiency better. The results indicate that cryptocurrencies are inefficient show strong signs of nonlinear serial dependence. Most of the cryptocurrencies exhibit volatility clusters, deterministic chaos, and long-range autocorrelation. These findings have useful implications for portfolio managers and other market players.
  Keywords: cryptocurrencies, weak-form efficient market hypothesis, detrended fluctuation analysis, mutual information, Lyapunov exponents
  

• Xavier, António (CEFAGE-UE: Center for Advanced Studies in Management and Economics); Fragoso, Rui (CEFAGE-UE: Center for Advanced Studies in Management and Economics, Management Department, Universidade de Évora, Évora, Portugal); Tomás, José Carlos (Direção Regional de Agricultura e Pescas do Algarve); Ludovico, Carlos (Direção Regional de Agricultura e Pescas do Algarve
  An approach to disaggregate the citrus crop at a local level using entropy and supervised classifications
  The changes of the Common Agricultural Policy (CAP) have led to several consequences in the land-uses and, in the environment, which call for disaggregated agricultural data at a local level. In order to tackle such problem, data disaggregation approaches are needed. Previous studies, carried out by the authors, combined supervised classifications and the LUCAS survey with an entropy approach to disaggregate agricultural data at a local level. This approach has two steps: 1) a supervised classification is developed using satellite imagery and the LUCAS survey; 2) an entropy approach is used to guaranty consistence among several sources of information and the aggregate. However, there are several investigation lines that weren’t still developed by the authors 1) The approach wasn’t implemented using a finer grid, which presents a challenge due to the computational burden 2) The approach may be modified to calculate the aggregate and the real areas, providing a way of comparison with the areas of the official statistics. This is special important due to the discussion being carried out regarding the national agricultural census data. Therefore, in the Algarve disaggregated recent data is needed at a more detailed grid, namely for the citrus crops. This paper continues this investigation and proposes an approach estimate recent data for the citrus crop at a very detailed pixel level and to calculate the total area for the Algarve region. The model disaggregated data into a 16 and 25 hectares pixel grid for the Algarve region, considering, therefore, more than 35.000 disaggregated units. Provisional results a with a 1 ha grid are also presented for the Silves municipality, considering more than 70.000 disaggregated units. The results were validated and showed a good correlation between observed and estimated land-uses, and may be of value for public institutions.
  Keywords: data disaggregation; supervised classifications; classification algorithms; minimum cross-entropy; LANDSAT 8, SENTINEL 2A, citrus crop, Algarve
  

• Carvalho, João (IEETA/DETI/UA); Brás, Susana (IEETA/DETI/UA); Pinho, Armando J. (IEETA/DETI/UA)
  ECG biometric identification using relative compression
  To understand pathological characteristics, in clinical practice, it is usual to try to reduce the inter-variability that characterizes the ECG signal. This inter-variability is precisely the source of richness that renders the ECG an interesting signal for biometric applications. Due to its characteristics (universality, uniqueness, measurability, acceptability and circumvention avoidance), there is a trend in biometrics to use the ECG signal for personal identification. Recent works based on compression models have shown that these approaches are suitable to ECG biometric identification. However, the best results are usually achieved by fiducial methods -- methods that rely on the detection of at least a fiducial point in each heartbeat found in the signal. In this work, we propose a compression-based non-fiducial method, that uses a measure of similarity, called the Normalized Relative Compression--a measure that tries to approach the Kolmogorov complexity of strings. Our method uses extended-alphabet finite-context models on the quantized first-order derivative of the signal, instead of using directly the original signal, as other methods do. We were able to achieve state-of-the-art results on a database collected at the University of Aveiro, which was used on previous works, making it a good preliminary benchmark for the method.
  Keywords: kolmogorov complexity, signal processing, compression metrics, ecg, biometrics
  

• Pratas, Diogo (IEETA/DETI/UA)
  Evolutionary patterns in coronaviruses genomes
  Advances in sequencing technologies have enabled the characterization of multiple microbial and host genomes, opening new frontiers of knowledge while kindling novel applications and research perspectives. Among these is the investigation of the viral communities residing or infecting the human body and their impact on health and disease. This talk will provide a comprehensive pattern analysis of some of the most popular viral genomes, including coronaviruses, using data compression and relative singularity analysis. We will provide the localization of minimal relative absent words (mRAW) as the shortest DNA (or RNA) sequences that exist in a pathogen and are absent from its host DNA, expressing pathogen signatures with the potential to build fast diagnostic and targeted therapeutics. We will provide additional characteristics and singularities of SARS-CoV-2, namely through comparisons with the existing similar viruses using time dynamics and data complexity analysis, rendering insights into the evolutionary nature of the current Covid-19 pandemic.
  Keywords: SARS-CoV-2, relative singularity, data compression, Kolmogorov complexity
  

• Coelho, Bruno (Instituto de Telecomunicações); Rocha-Pinto, Hélio (Universidade Federal do Rio de Janeiro, Observatório Nacional); Guedes,Leandro (Planetarium Foundation of the City of Rio de Janeiro); Andrei, Alexandre (Universidade Federal do Rio de Janeiro, Observatório do Valongo; Observatório Nacional- MCTIC; SYRTE, Observatoire de Paris); Lyra, Alexandre (Universidade Federal do Rio de Janeiro, História das Ciências das Tècnicas e Epistemologia, Observatório do Valongo); Bedin, Luciano (Departamento de Matemática - Universidade de Santa Catarina); Mattos Antunes, Marcelo (Secretaria de Estado de Educação); Rego, Elias (Universidade Federal Rio de Janeiro, Núcleo de Ciências Exactas)
  Defining the hyper metallicity stars count through MaxEnt statistics
  Through nuclear fusion of hydrogen in their cores, stars synthesize heavier elements from helium to iron. The rate of abundance from iron to hydrogen in stars is termed metallicity. It mainly depends of the stellar ages and conditions of their places of birth. Metallicity measurements are not so simple; in general the samples are deficient in subjects with respect to the total number of stars of the Galaxy. The Method of Maximum Entropy (MaxEnt; Jaynes, 1957) thus represents the indicate statistics to handle and interpret such samples (see Andrei et al, 2019). Here we use the data from the Geneva-Copenhagen survey of the Solar neighbourhood (Nordström et al., 2004), which brings ages, metallicities, and kinematics properties of nearly 13,500 dwarf stars of spectral type F and G analogous to the Sun. On our initial step we analyzed the Z velocity component perpendicular to the thin disk to assert that the number of high velocity stars was small indicating no relevant quantity of contaminants from outside the external regions. Next, we proceeded to study the ultra metal rich [Fe/H] > 0.5 and metal deficient [Fe/H] < -0.5 stars. By employing the MaxEnt we were able to rigorously consider these extreme cases and hence to derive a most detailed metallicity function of solar type stars in the thin disk. Three results are highlighted here. First, the irrelevance of extreme metal poor stars, even the oldest. Second, on how the MaxEnt is so efficient on cleansing the spurious contribution of old stars towards the galactic center over the counts of the extreme metal rich stars. And finally, our end goal, a robust determination of the extreme metal rich stars metallicity function.
  Keywords: maximum entropy, astrophysics, galaxy, metallicity
  

• Angelelli, Mario (Department of Innovation Engineering and CAMPI lab, University of Salento; INdAM; INFN); Ciavolino, Enrico (Department of History, Society and Human Study, University of Salento)
  Algebraic structures for knowledge representation: information-theoretic properties of determinantal expansions and applications
  Several models in applied physical and social science rely on structures defined by equations over a given field: depending on the context, for complex systems such equations may represent constraints that connect observed quantities (e.g. extensive quantities in statistical physics, manifest variables for factorial analysis in psychometry) with theoretical constructs or proxy variables. Maximum entropy is a fundamental method to represent our knowledge of such systems without adding further information beyond these constraints.
  This contributions addresses the problem of exploring the information content of such constraints when they are expressed in algebraic form. In particular, we focus on a special class of equations arising from determinantal expansions. This type of equations provide an algebraic basis for the description of (in-)dependence relations, both qualitatively (i.e. combinatorially) and quantitatively (i.e. in statistical terms). This contributions aims at exploring the algebro-geometric properties of determinantal expansions in an information-theoretic in order to provide a basis for future applications in the study of relations induced by subsets of variables, with particular regard to partial correlations.
  We provide new results regarding the complexity reduction for sign configurations coming from determinantal constraints. Such complexity reduction is quantified through the relative entropy associated with the check of the constraint satisfaction. A connection with the theory of integrable systems is drawn. A geometric interpretation and its applications in knowledge representation and psychometry are discussed too.
  Keywords: dependence structures, complexity reduction, determinantal expansion, relative entropy, partial correlations
  

• Ferreira, Paulo (VALORIZA, CEFAGE - IPP); Almeida, Dora (CEFAGE - Universidade de Évora); Dionísio, Andreia (CEFAGE - Universidade de Évora); Bouri, Ellie (Lebanese American University); Quintino, Derick (University of São Paulo)
  Energy markets – who are the influencers?
  The energy markets have recently undergone important transformations and witnessed a variety of crisis periods that have affected the relationships among energy commodities and their interactions with clean energy stocks. This has led to implications for price discovery, asset allocation, and risk management. It is thus important to uncover and identify which energy indices (or forms of energy) lead others or are the most influential. To uncover the complex structure of the relationship across the returns of seven different energy commodities and two clean energy stock indices, we apply Granger causality and transfer entropy in both static and dynamic approaches. The Granger causality analysis identifies the influence of the other energy products on natural gas, the transfer entropy reveals not only the importance of WTI oil but also the influence of clean energy indices. Diesel is the most influenced energy commodity. A rolling windows analysis confirms those findings, although it shows evidence of a time-variation that reflects the impacts of crisis periods, especially the pandemic, on the dynamics of relationships.
  Keywords: Energy markets; Granger causality; Transfer Entropy; Influencer
  

• Wu, Ximing (Department of Agricultural Economics, Texas A&M University)
  Maximum entropy density-quantile estimator for quantile-grouped data
  The Maximum Entropy (ME) distribution for a frequency table of intervals with known boundaries is known to be the histogram. The study proposes its counterpart for averages of intervals defined by sample quantiles. This is a difficult problem when the interval boundaries are not reported. We transform both the differential entropy and the interval averages into functions of the density-quantile of the underlying distribution and show that the Maximum Entropy principle yields a unique ME density-quantile function. Based on this ME construct, we derive analytical results for the associated quantile, distribution, density, Lorenz curve and Gini index. This procedure is further augmented to accommodate definite boundary conditions. We present a number of illustrations and an application to estimating annual Brazilian income distributions. Lastly we explore the informativeness of equal-space and equal-frequency sample partition based on their ME distributions. Our results suggest that equal-frequency partition adapts to the underlying distribution and can be more informative. The proposed estimators provide an approach to construct distributions based on quantile-grouped summaries. This approach is simple to implement, numerically stable, uniquely defined by the given interval averages and free of tuning parameters.
  Keywords: maximum entropy; interval summary; quantile; density
  

• Guiomar, Fernando (Instituto de Telecomunicações); Monteiro, Paulo (Instituto de Telecomnicações)
  Maximizing information capacity of optical communication systems through probabilistic constellation shaping
  Leveraged by an exponential increase of bandwidth demand, strongly fostered by a new revolution on the telecom industry with the upcoming 5G era, optical fiber systems are progressively reaching its capacity limits. This challenge calls for the development of novel techniques to enable an efficient modulation and transmission of the optical signal over the fiber channel. Recently, there have been notable advances on the application of modern information theoretic concepts to high-capacity optical fiber systems, namely including advanced modulation and coding. In this presentation, we will revise the recent developments on probabilistic constellation shaping techniques, aiming at closing the gap between practical optical transmission systems and Shannon's channel capacity. We will also address the problem of entropy maximization in such systems, which is tightly associated with the maximization of data rate for a given communication channel. Key practical issues such as the design of distribution matching algorithms, adaptability to time- and frequency-varying conditions, and application to realistic network scenarios, will also be addressed and backed up by experimental analyses. Finally, we will discuss a few open problems that still require an in-depth mathematical analysis for their appropriate treatment and therefore remain to be solved by a research community that is strongly dominated by pragmatic application-oriented engineers.
  Keywords: optical communications, information theory, coding, shaping, modulation