No abstract

No abstract

The optimization of expensive to evaluate, black-box, mixed-variable functions, i.e. functions that have continuous and discrete inputs, is a difficult and yet pervasive problem in science and engineering. In Bayesian optimization (BO), special cases of this problem that consider fully continuous or fully discrete domains have been widely studied. However, few methods exist for mixed-variable domains and none of them can handle discrete constraints that arise in many real-world applications. In this paper, we introduce MiVaBo, a novel BO algorithm for the efficient optimization of mixed-variable functions combining a linear surrogate model based on expressive feature representations with Thompson sampling. We propose an effective method to optimize its acquisition function, a challenging problem for mixed-variable domains, making MiVaBo the first BO method that can handle complex constraints over the discrete variables. Moreover, we provide the first convergence analysis of a mixed-variable BO algorithm. Finally, we show that MiVaBo is significantly more sample efficient than state-of-the-art mixed-variable BO algorithms on several hyperparameter tuning tasks, including the tuning of deep generative models.

This article presents a thorough investigation of the five Russian deictic words that correspond to the English meanings 'here ' and 'now': zdes', tut, sejčas, teper' and vot. We analyze data from the Russian National Corpus and data from Russian TV news broadcasts. On the basis of the corpus data, we propose a radial category network consisting of nine subcategories, which encompass all five words, and show that although the deictic words have overlapping distributions, they all have distinct "radial category profiles" in the sense that they display different centers of gravity in the network. We advance the "Minimal Adaptation Hypothesis", according to which language makes adaptations that are as small as possible, when applied to a new setting, such as the one created by TV.Аннотация: В статье представлено тщательное исследование пяти дейктических слов русского языка, соответсвующих английским местоимениям 'here' и 'now': здесь, тут, сейчас, теперь и вот. Мы проанализировали данные Национального корпуса русского языка, а также данные русских телевизионных выпусков новостей. На основе данных корпуса мы предлагаем радиальную сеть значений, состоящую из девяти подкатегорий и описывающую все пять дейктических слов. Мы показываем, что, хотя исследуемые дейктические единицы пересекаются в их распределении, каждая из них имеет свой «профиль радиальной категории», иными словами, свой центр тяжести в рамках предложенной радиальной сети значений. Мы также выдвигаем «гипотезу минимальной адаптации», согласно которой приспособление языка к новым условиям функционирования, таким как условия телевидения, оказывается по возможности минимальным.

No abstract

No abstract

This article presents a corpus-based investigation of temporal adverbials with special focus on Russian в 'in(to)' and its cognates in North Slavic (Belarusian, Ukrainian, Polish and Czech). We advance the Constraint Hypothesis, according to which case government is more restricted in the domain of time than in the domain of space. This hypothesis receives support from the five languages under scrutiny insofar as the distribution of the accusative vs. locative after в and its cognates is contrastive in the domain of space, but complementary in temporal adverbials. On this basis, we argue that the relationship between space and time is asymmetrical. Although all five languages display space-time asymmetries, we show that they have different systems of temporal adverbials, including a number of constructions with prepositions or bare cases. In order to capture the differences we propose a North Slavic Temporal Adverbial Continuum, which is corroborated by a thorough statistical investigation of data from the ParaSol corpus.Аннотация: В статье представлено корпусное исследование обстоятельств времени в северных славянских языках (русском, белорусском, украинском, польском и чешском). В фокусе внимания данной работы русские обстоятельства времени с предлогом в и их соответствия в северных славянских языках. Мы предлагаем Гипотезу Ограничения, согласно которой падежное управление менее свободно в сфере выражения временных отношений, чем пространственных. Эта гипотеза находит подтверждение на материале пяти славянских языков: распределение винительного и предложного падежей после предлога в (и его аналогов в других славянских языках) является контрастным в пространственных, и дополнительным во временных конструкциях. На этом основании мы утверждаем, что отношения между выражением пространства и времени в языке являются асимметричными. Такая пространственно-временная асимметричность наблюдается во всех пяти исследуемых языках, однако системы обстоятельств времени в этих языка различны. В функции обстоятельств времени используются как предложно-падежные конструкции, так и падежные формы без предлогов. Разнообразие в способах оформления обстоятельств времени в северных славянских языках позволяет постулировать Северно-славянский Континуум Обстоятельств Времени. Выдвинутые в работе гипотезы подтверждаются статистическим анализом данных из корпуса ParaSol.

Nearly two thousand perfective verbs in Russian are formed via the addition of so-called "empty prefixes" (čistovidovye pristavki) to imperfective base verbs. The traditional assumption that prefixes are semantically "empty" when used to form aspectual pairs is problematic because the same prefixes are clearly "nonempty" when combined with other base verbs. Though some scholars have suspected that the prefixes are not empty but instead have meanings that overlap with the meanings of the base verbs, proof of this hypothesis has eluded researchers. With the advent of corpora and electronic resources it is possible to explore this question on the basis of large quantities of data. This article presents a new methodology, called "Radial category profiling", in which the semantic network of a prefix is established on the basis of its "non-empty" uses and then compared, node by node, with the semantic network of base verbs that use the same prefix as an "empty" perfectivizing morpheme. This methodology facilitates a comprehensive analysis of ten prefixes, comparing their meanings in "nonempty" and "empty" uses and showing precisely how in the latter case overlap produces the illusion of emptiness. We are able to fully specify the semantic network of each prefix, and discover that for some prefixes there is overlap througout the network, while for others overlap is restricted to a contiguous subsection of the network. We investigate the dynamic interactions among prefixes, and identify what meanings are incompatible with the "purely aspectual" function of the so-called "empty" prefixes.