A. Baklanov

Atmospheric transport patterns and possible consequences for the European North after a nuclear accident.

The main purpose of this study is to examine possible impacts and consequences of a hypothetical accident at the Kola nuclear plant in north-west Russia on different geographical regions: Scandinavia, central Europe, European FSU and Taymyr. The period studied is 1991-1996. An isentropic trajectory model has been used to calculate forward trajectories that originated over the nuclear accident region. Atmospheric transport patterns were identified using the isentropic trajectories and a cluster analysis technique. From the trajectory model results, a number of cases were chosen for examination in detail using more complete transport models. For this purpose, the models MATHEW/ADPIC, DERMA and a newly developed FOA Random Displacement Model have been used to simulate the radionuclide transport and contamination in the case of a nuclear accident and their results have been compared with those of the trajectory modelling. Estimation of the long-term consequences for populations after an accident has been performed for several specific dates by empirical models and correlation between fallout and doses to humans on the basis of the Chernobyl accident exposures in Scandinavia.

Volatility Prediction using Financial Disclosures Sentiments with Word Embedding-based IR Models

Volatility prediction--an essential concept in financial markets--has recently been addressed using sentiment analysis methods. We investigate the sentiment of annual disclosures of companies in stock markets to forecast volatility. We specifically explore the use of recent Information Retrieval (IR) term weighting models that are effectively extended by related terms using word embeddings. In parallel to textual information, factual market data have been widely used as the mainstream approach to forecast market risk. We therefore study different fusion methods to combine text and market data resources. Our word embedding-based approach significantly outperforms state-of-the-art methods. In addition, we investigate the characteristics of the reports of the companies in different financial sectors.

Vote Aggregation Techniques in the Geo-Wiki Crowdsourcing Game: A Case Study

The Cropland Capture game (CCG) aims to map cultivated lands using around 170000 satellite images. The contribution of the paper is threefold: (a) we improve the quality of the CCG’s dataset, (b) we benchmark state-of-the-art algorithms designed for an aggregation of votes in a crowdsourcing-like setting and compare the results with machine learning algorithms, (c) we propose an explanation for surprisingly similar accuracy of all examined algorithms. To accomplish (a), we detect image duplicates using the perceptual hash function pHash. In addition, using a blur detection algorithm, we filter out unidentifiable images. In part (c), we suggest that if all workers are accurate, the task assignment in the dataset is highly irregular, then state-of-the-art algorithms perform on a par with Majority Voting. We increase the estimated consistency with expert opinions from 77% to 91% and up to 96% if we restrict our attention to images with more than 9 votes.

On the question of construction of an attraction set under constraints of asymptotic nature

We study a variant of the reachability problem with constraints of asymptotic character on the choice of controls. More exactly, we consider a control problem in the class of impulses of given intensity and vanishingly small length. The situation is complicated by the presence of discontinuous dependences, which produce effects of the type of multiplying a discontinuous function by a generalized function. The constructed extensions in the special class of finitely additive measures make it possible to present the required solution, defined as an asymptotic analog of a reachable set, in terms of a continuous image of a compact, which is described with the use of the Stone space corresponding to the natural algebra of sets of the control interval.One of the authors had the honor of communicating with Nikolai Nikolaevich Krasovskii for many years and discussed with him problems that led to the statement considered in the paper. Krasovskii’s support of this research direction provided possibilities for its fruitful development. His disciples and colleagues will always cherish the memory of Nikolai Nikolaevich in their hearts.

Demand side management: A case for disruptive behaviour

The UK electricity system is undergoing a significant transformation. Increasing penetration of renewable generation and integration of new consumer technologies (e.g. electric vehicles) challenge the traditional way of balancing electricity in the grid, whereby supply matches demand. Demand-side management (DSM) has been shown to offer a promising solution to the above problem. However, models proposed in literature typically consider an isolated system whereby a single aggregator coordinates homogeneous consumers. As a result potential externalities of DSM are overlooked. This work explores the value of DSM in the context of an interacting electricity system, where utilities compete for cheap electricity in the wholesale market. A stylized model of the UK electricity system is proposed, whereby a traditional supplier competes with a ‘green’ supplier in the wholesale market. The modelling was able to show that with enough dispatchable capacity the traditional supplier was able to benefit from instructing his consumers to increase demand peaks, which had an adverse effect on the system.

On a density property of weakly absolutely continuous measures. General case

It is shown that some set of all step functions (and the set of all uniform limits of such functions) allows an embedding into a compact subset (with respect to weak-star topology) of the set of all finitely additive measures of bounded variation in the form of an everywhere dense subset. In particular, we consider the set of all step functions (the set of all uniform limits of such functions) such that an integral of absolute value of the functions with respect to nonnegative finitely additive measure λ is equal to unity. For these sets, the possibility of embedding is proved without any additional assumptions on λ; this generalizes the previous results. Using the Sobczyk-Hammer decomposition theorem, we show that for λ with the finite range, the above-mentioned sets of functions allow an embedding into the unit sphere (in the strong norm-variation) of weakly absolutely continuous measures with respect to λ in the form of an everywhere dense subset. For λ with an infinite range, the above-mentioned sets of functions allow an embedding into the unit ball of weakly absolutely continuous measures with respect to λ in the form of an everywhere dense subset. The results can be helpful for an extension of linear impulse control problems in the class of finitely additive measures to obtain robust representations of reachable sets given by constraints of asymptotic character.

The Cropland Capture Game: good annotators versus vote aggregation methods

The Cropland Capture game, which is a recently developed Geo-Wiki game, aims to map cultivated lands using around 17,000 satellite images from the Earth’s surface. Using a perceptual hash and blur detection algorithm, we improve the quality of the Cropland Capture game’s dataset. We then benchmark state-of-the-art algorithms for an aggregation of votes using results of well-known machine learning algorithms as a baseline. We demonstrate that volunteer-image assignment is highly irregular and only good annotators are presented (there are no spammers and malicious voters). We conjecture that the last fact is the main reason for surprisingly similar accuracy levels across all examined algorithms. Finally, we increase the estimated consistency with expert opinion from 77 to 91 % and up to 96 % if we restrict our attention to images with more than 9 votes.

On an asymptotic analysis problem related to the construction of an attainability domain

Problems of constructing and analyzing the properties of attainability domains play an important role in control theory and its applications. In particular, this applies to control under impulse constraints that reflect the energetics of a process. The situation is complicated by the possible instability of the process under variation (in particular, under relaxation) of constraints related to boundary and intermediate conditions. Stability of the problem is also missing, in general, under relaxation of state constraints. In these cases, it is natural to focus on the asymptotic variant of the statement; this is especially expedient when one has to deal with initially asymptotic requirements. In all these cases, it seems expedient to use analogs of J. Warga’s approximate solutions. At the same time, to seek necessary approximate (and, in fact, asymptotic) solutions, it is natural to use generalized modes. For problems with impulse constraints and discontinuity in the coefficients of control actions, such modes lead to phenomena described by products of discontinuous functions and generalized functions even in the class of linear systems. In a large series of his studies, to overcome the arising difficulties, one of the authors used constructions of extension in the class of finitely additive measures. The present paper follows this approach and is ideologically relevant to the engineering problem of controlling the thrust of an engine under conditions of a given program of variation of its orientation; it is postulated that energy resources are completely consumed in a natural (for a number of impulse control problems) mode of short-duration impulses: the set of time instants at which the instantaneous control is different from zero can be embedded in an interval of vanishingly small length. Within these short periods of time, the engine should consume all energy resources while obeying some other constraints (making the sense of moment constraints) to a high degree of accuracy. In addition, one should take into account the possible discontinuity of the functions defining the coefficients of control actions. As a natural analog of the attainability domain, we use an attraction set, whose construction, together with the subsequent study of its main properties, constitutes the goal of the present study.