Secure Aggregation protocols allow a collection of mutually distrust parties, each holding private value, to collaboratively compute the sum those values without revealing themselves. We consider training deep neural network in Federated Learning model, using distributed stochastic gradient descent across user-held data on mobile devices, wherein protects user's model gradient. design novel, communication-efficient protocol for high-dimensional that tolerates up 1/3 users failing complete...

In this paper we explore to what extent text parameters, such as average number of words per sentence, syllables word, nouns frequency content words, etc. can successfully rank Russian academic texts for different age and grade levels. We provide a brief overview previous re search on readability describe the corpus school textbooks Social Studies (from 5-th 11-th grade) compiled by authors. share our experience using variety quantitative complexity metrics evaluate measures existing...

It is a cliche to say that there gap between research and practice. As the interest importance in practical impact of has been growing, practice expected be narrowing. However, our study reveals still seems wide gap. We survey so ware engineers about what they care when developing ware. then compare results with topics papers published ICSE/FSE recently. found following discrepancy: while more development productivity than quality ware, on areas closely related productivity--such as process...

This paper presents action research to analyze an approach for assessment of the alleged agile transformation. was implemented at AK Bars Digital Technologies, IT spin-off one largest banks in Russia using Scaled Agile Framework. The is based on Goal-Question-Metric approach, non-invasive measurement collection, and systemic analysis. It uses data from several different sources, including interviews, code repositories, user ratings play stores, templates assessment. effectiveness...

BACKGROUND: The low solubility of pinostrobin in water limits its use, causing technological difficulties and significantly reducing bioavailability. "solid dispersion method" can increase both the release a number active substances from various dosage forms. AIM: purpose this study is pharmacological solid based on it. MATERIAL AND METHODS: Experiments were performed vitro vivo cells human hepatocellular carcinoma (HepG2) 90 white outbred male female rats 24 mice. Methods used to assess...

In this paper, we present a shared task on core information extraction problems, named entity recognition and relation extraction. contrast to popular tasks related try move away from strictly academic rigor rather model business case. As source for textual data choose the corpus of Russian strategic documents, which annotated according our own annotation scheme. To speed up process, exploit various active learning techniques. total ended with more than two hundred documents. Thus managed...

Education policy makers view measuring academic texts readability and profiling classroom textbooks as a primary task of education management aimed at sustaining quality reading programs. As Russian metrics, i.e. "objective" features determining its complexity for readers, re still research niche, we undertook comparative analysis exemplified in on Social Science examination foreign language. Experiments 7 classifiers 4 methods linear regression Readability corpus demonstrated that ranking...

Abstract Motivation This article describes NEREL-BIO—an annotation scheme and corpus of PubMed abstracts in Russian smaller number English. NEREL-BIO extends the general domain dataset NEREL by introducing domain-specific entity types. covers both biomedical domains making it suitable for transfer experiments. provides nested named entities as an extension employed NEREL. Nested may cross boundaries to connect shorter within longer entities, them harder detect. Results contains annotations...

In this paper, we present NEREL, a Russian dataset for named entity recognition and relation extraction.NEREL is significantly larger than existing datasets: to date it contains 56K annotated entities 39K relations.Its important difference from previous datasets annotation of nested entities, as well relations within at the discourse level.NEREL can facilitate development novel models that extract between on both sentence document levels.NEREL also events involving their roles in events.The...

Automatic event extraction form text is an important step in knowledge acquisition and base population. Manual work development of system indispensable either corpus annotation or vocabularies pattern creation for a knowledge-based system. Recent works have been focused on adaptation existing (for from English texts) to new domains. Event other languages was not studied due the lack resources algorithms necessary natural language processing. In this paper we define set linguistic that are...

Despite increasing popularity of developer dashboards, the effectiveness dashboards is still in question. In order to design a dashboard that effective and useful for developers, it important know (a) what information developers need see dashboard, (b) how want use with necessary information. To answer these questions, we conducted two series face-to-face individual interviews developers. first step analyzed answers, build Goal-Question-Metric model designed precooked dashboard. Then, during...

The problem of sentiment analysis has been widely studied in the past several decades. research area predominantly based on data collated from online messages, microblogs, reviews, etc. Significantly fewer studies have conducted academic discourse and especially school textbooks. However, texts can help answer pressing issues relating ways which different referents are presented contemporary Russian In this paper, we analyze distribution words phrases a Corpus textbooks History (Grades...

Software is mostly, if not entirely, a knowledge artifact. best practices are often thought to work because they induce more productive behaviour in software developers. In this paper we deployed new generation tool, portable multichannel EEG, obtain direct physical insight into the mental processes of working developers engaged their standard activities. We have demonstrated feasibility approach and obtained glimpse its potential power distinguish brain activity with different methodologies.