Search

Deep learning techniques are required for the analysis of synoptic (multi-band and multi-epoch) light curves in massive data of quasars, as expected from the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST). In this follow-up study, we introduce an upgraded version of a conditional neural process (CNP) embedded in a multi-step approach for the analysis of large data of quasars in the LSST Active Galactic Nuclei Scientific Collaboration data challenge database. We present a case study of a stratified set of u-band light curves for 283 quasars with very low variability ∼0.03. In this sample, the CNP average mean square error is found to be ∼5% (∼0.5 mag). Interestingly, besides similar levels of variability, there are indications that individual light curves show flare-like features. According to the preliminary structure–function analysis, these occurrences may be associated with microlensing events with larger time scales of 5–10 years.

Gaussian conditional random fields (GCRF) are a well-known used structured model for continuous outputs that uses multiple unstructured predictors to form its features and at the same time exploits dependence structure among outputs, which is provided by a similarity measure. In this paper, a Gaussian conditional random fields model for structured binary classification (GCRFBC) is proposed. The model is applicable to classification problems with undirected graphs, intractable for standard classification CRFs. The model representation of GCRFBC is extended by latent variables which yield some appealing properties. Thanks to the GCRF latent structure, the model becomes tractable, efficient and open to improvements previously applied to GCRF regression models. In addition, the model allows for reduction of noise, that might appear if structures were defined directly between discrete outputs. Additionally, two different forms of the algorithm are presented: GCRFBCb (GCRGBC - Bayesian) and GCRFBCnb (GCRFBC - non Bayesian). The extended method of local variational approximation of sigmoid function is used for solving empirical Bayes in Bayesian GCRFBCb variant, whereas MAP value of latent variables is the basis for learning and inference in the GCRFBCnb variant. The inference in GCRFBCb is solved by Newton-Cotes formulas for one-dimensional integration. Both models are evaluated on synthetic data and real-world data. It was shown that both models achieve better prediction performance than unstructured predictors. Furthermore, computational and memory complexity is evaluated. Advantages and disadvantages of the proposed GCRFBCb and GCRFBCnb are discussed in detail., Draft paper without experimental evaluation

In this paper, a new pseudo-random number generator (PRNG) based on improved onedimensional discrete-space chaotic map is proposed. Like the original, the improved map relies on bijective mapping of permutations and natural numbers. Instead of using standard Lehmer code, we use a mapping computable in linear time, which significantly speeds up the PRNG. Results of NIST 800-22 test suite and TestU01 test suite confirm that the proposed approach can be used for generation of pseudo-random numbers. Due to discrete nature of used chaotic map, the proposed PRNG is not influenced by dynamical degradation and has virtually unlimited key space. Proposed approach has much better ratio between required memory and security level than previous secure one-dimensional discrete-space chaotic PRNGs. Also, proposed PRNG is much faster than other secure PRNGs of the same type. Satisfactory speed and small memory requirements indicate that proposed PRNG has properties desirable for use in devices with limited memory space, such as wireless sensor networks.

Regression kriging is one of the most popular spatial interpolation techniques. Its main strength is that it exploits both spatial autocorrelation as well as information contained in environmental covariates. While regression kriging is still dominant, in the past few years machine learning, especially random forest, is increasingly being used for mapping. Machine learning is more flexible than multiple linear regression and can thus make better use of environmental covariates. But machine learning typically ignores spatial autocorrelation. Several attempts have been made to include spatial autocorrelation in random forest, by adding distances to observation locations and other geometries to the set of covariates. But none of these studies have tried the obvious solution to include the nearest observations themselves and the distances to the nearest observations as covariates. In this study we tried this solution by introducing and testing Random Forest for Spatial Interpolation (RFSI). RFSI trains a random forest model on environmental covariates as well as nearest observations and their distances from the prediction point. We applied and evaluated RFSI for mapping daily precipitation in Catalonia for the 2016-2018 period. We trained four models (random forests, RFsp, pooled regression kriging and RFSI) using 63,927 daily precipitation observations from 87 GHCN-stations located in Catalonia. Maximum and minimum daily temperatures and IMERG precipitation estimates (derived from the GPM mission) were used as environmental covariates for all four models. Results based on 5-fold cross validation showed that RFSI (R-square 69.4%, RMSE 3.8 mm) significantly outperformed all random forest (R-square 50.6%, RMSE 3.8 mm), RFsp (R-square 55.5%, RMSE 4.6 mm) and pooled regression kriging (R-square 65.3%, RMSE 4.0 mm). Finetuning RFSI could potentially improve prediction accuracy even more. In addition to improved prediction accuracy, RFSI has the advantage that it uses much fewer spatial covariates than RFsp.

In recent years, portfolio problem solving found many applications in automated reasoning, primarily in SAT solving and in automated and interactive theorem proving. Portfolio problem solving is an approach in which for an individual instance of a specific problem, one particular, hopefully most appropriate, solving technique is automatically selected among several available ones and used. The selection usually employs machine learning methods. To our knowledge, this approach has not been used in automated theorem proving in geometry so far and it poses a number of new challenges. In this paper we propose a set of features which characterize a specific geometric theorem, so that machine learning techniques can be used in geometry. Relying on these features and using different machine learning techniques, we constructed several portfolios for theorem proving in geometry and also runtime prediction models for provers involved. The evaluation was performed on two corpora of geometric theorems: one coming from geometric construction problems and one from a benchmark set of the GeoGebra tool. The obtained results show that machine learning techniques can be useful in automated theorem proving in geometry, while there is still room for further progress.

An approach for using lasso (Least Absolute Shrinkage and Selection Operator) regression in creating sparse 3D models of soil properties for spatial prediction at multiple depths is presented. Modeling soil properties in 3D benefits from interactions of spatial predictors with soil depth and its polynomial expansion, which yields a large number of model variables (and corresponding model parameters). Lasso is able to perform variable selection, hence reducing the number of model parameters and making the model more easily interpretable. This also prevents overfitting, which makes the model more accurate. The presented approach was tested using four variable selection approaches – none, stepwise, lasso and hierarchical lasso, on four kinds of models – standard linear model, linear model with polynomial expansion of depth, linear model with interactions of covariates with depth and linear model with interactions of covariates with depth and its polynomial expansion. This framework was used to predict Soil Organic Carbon (SOC) in three contrasting study areas: Bor (Serbia), Edgeroi (Australia) and the Netherlands. Results show that lasso yields substantial improvements in accuracy over standard and stepwise regression — up to 50 % of total variance. It yields models which contain up to five times less nonzero parameters than the full models and that are usually more sparse than models obtained by stepwise regression, up to three times. Extension of the standard linear model by including interactions typically improves the accuracy of models produced by lasso, but is detrimental to standard and stepwise regression. Regarding computation time, it was demonstrated that lasso is several orders of magnitude more efficient than stepwise regression for models with tens or hundreds of variables (including interactions). Proper model evaluation is emphasized. Considering the fact that lasso requires meta-parameter tuning, standard cross-validation does not suffice for adequate model evaluation, hence a nested cross-validation was employed. The presented approach is implemented as publicly available sparsereg3D R package.

Artificial intelligence is steadily increasing its impact on everyday life. Therefore, the societal issues of artificial intelligence have become an important concern in the AI research. The presence of data that reflects human biases towards historically discriminated groups defined by sensitive features such as race and gender, results in machine learning models which discriminate against these groups. In order to tackle the impact of bias in data, researchers developed a variety of specialized machine learning algorithms which are able to satisfy different fairness constraints imposed on the model. Group fairness constraints do not fit standard machine learning formulations easily due to their non-differentiable nature. In this paper we developed a technique for learning a fair classifier by Monte Carlo policy gradient method which naturally deals with such non-differentiable constraints. Our methodology focuses on direct optimization of both group fairness metric and predictive performance of the model. In addition, we propose two different variance reduction techniques of gradient estimation. We compare our models to seven other related and state-of-the-art models and demonstrate that they are able to achieve better trade-off between accuracy and unfairness. To the best of our knowledge, this is the first fair classification algorithm which solves the issue of non-differentiable constraints by reinforcement learning techniques.

A new method for obtaining pseudo-random numbers, based on discrete-space chaotic map, is presented. The proposed method uses a discrete chaotic map, based on the composition of permutations. The randomness of pseudo-random sequences generated by proposed method is verified using NIST 800-22 test suite and TestU01. Proposed method is not affected by dynamical degradation, so the process of generation of pseudo-random numbers is not influenced by approximations of any kind. The advantage of the proposed method is that it has virtually unlimited key space and the ability to generate same number of different pseudo-random sequences as other secure discrete-space chaotic methods, but with significantly lower memory space requirements. Also, higher level of security and great cycle lengths can be achieved. Small memory requirements could make proposed PRNG applicable in devices with limited memory space.

For many decades, kriging and deterministic interpolation techniques, such as inverse distance weighting and nearest neighbour interpolation, have been the most popular spatial interpolation techniques. Kriging with external drift and regression kriging have become basic techniques that benefit both from spatial autocorrelation and covariate information. More recently, machine learning techniques, such as random forest and gradient boosting, have become increasingly popular and are now often used for spatial interpolation. Some attempts have been made to explicitly take the spatial component into account in machine learning, but so far, none of these approaches have taken the natural route of incorporating the nearest observations and their distances to the prediction location as covariates. In this research, we explored the value of including observations at the nearest locations and their distances from the prediction location by introducing Random Forest Spatial Interpolation (RFSI). We compared RFSI with deterministic interpolation methods, ordinary kriging, regression kriging, Random Forest and Random Forest for spatial prediction (RFsp) in three case studies. The first case study made use of synthetic data, i.e., simulations from normally distributed stationary random fields with a known semivariogram, for which ordinary kriging is known to be optimal. The second and third case studies evaluated the performance of the various interpolation methods using daily precipitation data for the 2016–2018 period in Catalonia, Spain, and mean daily temperature for the year 2008 in Croatia. Results of the synthetic case study showed that RFSI outperformed most simple deterministic interpolation techniques and had similar performance as inverse distance weighting and RFsp. As expected, kriging was the most accurate technique in the synthetic case study. In the precipitation and temperature case studies, RFSI mostly outperformed regression kriging, inverse distance weighting, random forest, and RFsp. Moreover, RFSI was substantially faster than RFsp, particularly when the training dataset was large and high-resolution prediction maps were made.

This article deals with teaching students how to review research articles of other authors and how to respond to reviewer comments related to their own articles. By teaching these skills, one educates students how to do something which is undoubtedly much more important for their professional development: how to properly do their own research and how to properly write their own research articles. The guidelines given are primarily intended for students of computer science. Nevertheless, the notions presented here are applicable to areas other than computing. Since there are so many branches within computing, each one characterized by its own particular features, this article also includes a section on variations of presented methodology, which are of interest for each particular branch of computer science covered by the authors of this article. In general, the review process consists of two main parts: formal review and quality review. Led by the lack of results in the available literature, an analysis based on readability index evaluates average article acceptance rates, review times, and delay period for publishing accepted articles depending on the following: article types, willingness of reviewers to do reviews, journal impact factor, and the scientific field. Quality review must be taken into account for all types of articles. Appendix A describes specific experiences from various scientific fields. Appendix B gives specific guidelines for a number of fields covered by the authors of this article.

This paper discusses research results with regard to energy-efficient transmission of serialised data between servers and mobile devices. A test environment was created in which the research authors primarily measured electricity consumption during communication between a mobile device and server. Numerical results were used to determine how well data serialisation was performed on a dedicated server and its effects on the power consumption of a mobile device. The time spent in data serialisation and the size of the serialised file were found to significantly influence energy consumption. Based on that fact, results have been used to create a mathematical model which was later introduced with functional forms. The main variables in those functional forms were time of serialisation and size of a serialised file. The data collected through this research has been used for an experimental API-CB Saver, which based on mathematical models chooses the most favourable manner of serialisation and compression in real time. The results collected during the tests show that the CBSaver-Api approach performs with greater energy efficiency than current techniques. Furthermore, with optimal selection of data serialisation type and compression level in real time the considered system shows better performance in power saving. According to the results, the API-CBSaver tests indicate the direction which one should take for the purposes of improving energy efficiency.

Evaluating improvements to modern SAT solvers and comparison of two arbitrary solvers is a challenging and important task. Relative performance of two solvers is usually assessed by running them on a set of SAT instances and comparing the number of solved instances and their running time in a straightforward manner. In this paper we point to shortcomings of this approach and advocate more reliable, statistically founded methodologies that could discriminate better between good and bad ideas. We present one such methodology and illustrate its application.

Context: The number of students enrolled in universities at standard and on-line programming courses is rapidly increasing. This calls for automated evaluation of students assignments. Objective: We aim to develop methods and tools for objective and reljable automated grading that can also provide substantial and comprehensible feedback. Our approach targets introductory programming courses, which have a number of specific features and goals. The benefits are twofold: reducing the workload for teachers, and providing helpful feedback to students in the process of learning. Method: For sophisticated automated evaluation of students' programs, our grading framework combines results of three approaches (i) testing, (ii) software verification, and (iii) control flow graph similarity measurement. We present our tools for software verification and control flow graph similarity measurement, which are publicly available and open source. The tools are based on an intermediate code representation, so they could be applied to a number of programming languages. Results: Empirical evaluation of the proposed grading framework is performed on a corpus of programs written by university students in programming language C within an introductory programming course. Results of the evaluation show that the synergy of proposed approaches improves the quality and precision of automated grading and that automatically generated grades are highly correlated with instructor-assigned grades. Also, the results show that our approach can be trained to adapt to teacher's grading style. Conclusions: In this paper we integrate several techniques for evaluation of student's assignments. The obtained results suggest that the presented tools can find real-world applications in automated grading. (C) 2012 Elsevier B.V. All rights reserved.

This paper proposes two compound measures of mapping quality to support objective comparison of spatial prediction techniques for geostatistical mapping: (1) mapping efficiency – defined as the costs per area per amount of variation explained by the model, and (2) information production efficiency – defined as the cost per byte of effective information produced. These were inspired by concepts of complexity from mathematics and physics. Complexity i.e. the total effective information is defined as bytes remaining after compression and after rounding up the numbers using half the mapping accuracy (effective precision). It is postulated that the mapping efficiency, for an area of given size and limited budget, is basically a function of inspection intensity and mapping accuracy. Both measures are illustrated using the Meuse and Ebergotzen case studies ( gstat , plotKML packages). The results demonstrate that, for mapping organic matter (Meuse data set), there is a gain in the mapping efficiency when using regression-kriging versus ordinary kriging: mapping efficiency is 7% better and the information production efficiency about 25% better (3.99 vs 3.14 EUR B−1 for the GZIP compression algorithm). For mapping sand content (Ebergotzen data set), the mapping efficiency for both ordinary kriging and regression-kriging is about the same; the information production efficiency is 29% better for regression-kriging (37.1 vs 27.7 EUR B−1 for the GZIP compression algorithm). Information production efficiency is possibly a more robust measure of mapping quality than mapping efficiency because: (1) it is scale-independent, (2) it can be more easily related to the concept of effective information content, and (3) it accounts for the extrapolation effects. The limitation of deriving the information production efficiency is that both reliable estimate of the model uncertainty and the mapping accuracy is required.

Demographic aspects of prehistoric populations have an important role in current archaeological theory and empirical research. In this study, we develop a method to estimate population dynamics and population size and apply it to data on house remains at one of key European Mesolithic-Neolithic transitional sites - Lepenski Vir (Serbia). Lepenski Vir is a site located in the Danube Gorges, well-known for its trapezoidal house floors and stone sculpture. It was most intensively occupied between similar to 6200 and similar to 6000 cal BC, the so called Transitional phase, which corresponds to the beginning of the Neolithic in Central Balkans. We combine archaeological evidence and ethnographic information with mathematical models of population dynamics and house accumulation within a Bayesian framework (Approximate Bayesian Computation) to derive posterior distributions of growth rate and population size estimates for the Lepenski Vir population in this period.

The solution of a class of third order ordinary differential equations possessing two parameter Lie symmetry group is obtained by group theoretic means. It is shown that reduction to quadratures is possible according to two scenarios: 1) if upon first reduction of order the obtained second order ordinary differential equation besides the inherited point symmetry acquires at least one more new point symmetry (possibly a hidden symmetry of Type II). 2) First, reduction paths of the fourth order differential equations with four parameter symmetry group leading to the first order equation possessing one known (inherited) symmetry are constructed. Then, reduction paths along which a third order equation possessing two-parameter symmetry group appears are singled out and followed until a first order equation possessing one known (inherited) symmetry are obtained. The method uses conditions for preservation, disappearance and reappearance of point symmetries., Comment: Part 2 of the paper consists of tables

We present a new, CDCL-based approach for automated theorem proving in coherent logic -- an expressive semi-decidable fragment of first-order logic that provides potential for obtaining human readable and machine verifiable proofs. The approach is described by means of an abstract state transition system, inspired by existing transition systems for SAT and represents its faithful lifting to coherent logic. The presented transition system includes techniques from which CDCL SAT solvers benefited the most (backjumping and lemma learning), but also allows generation of human readable proofs. In contrast to other approaches to theorem proving in coherent logic, reasoning involved need not to be ground. We prove key properties of the system, primarily that the system yields a semidecision procedure for coherent logic. As a consequence, the semidecidability of another fragment of first order logic which is a proper superset of coherent logic is also proven.

Execution of most of the modern DPLL-based SAT solvers is guided by a number of heuristics. Decisions made during the search process are usually driven by some fixed heuristic policies. Despite the outstanding progress in SAT solving in recent years, there is still an appealing lack of techniques for selecting policies appropriate for solving specific input formulae. In this paper we present a methodology for instance-based selection of solver's policies that uses a data-mining classification technique. The methodology also relies on analysis of relationships between formulae, their families, and their suitable solving strategies. The evaluation results are very good, demonstrate practical usability of the methodology, and encourage further efforts in this direction.