LIVIVO - The Search Portal for Life Sciences

zur deutschen Oberfläche wechseln
Advanced search

Search results

Result 1 - 10 of total 272

Search options

  1. Book ; Online: Probing Pretrained Models of Source Code

    Troshin, Sergey / Chirkova, Nadezhda

    2022  

    Abstract: Deep learning models are widely used for solving challenging code processing tasks, such as code generation or code summarization. Traditionally, a specific model architecture was carefully built to solve a particular code processing task. However, ... ...

    Abstract Deep learning models are widely used for solving challenging code processing tasks, such as code generation or code summarization. Traditionally, a specific model architecture was carefully built to solve a particular code processing task. However, recently general pretrained models such as CodeBERT or CodeT5 have been shown to outperform task-specific models in many applications. While pretrained models are known to learn complex patterns from data, they may fail to understand some properties of source code. To test diverse aspects of code understanding, we introduce a set of diagnosting probing tasks. We show that pretrained models of code indeed contain information about code syntactic structure and correctness, the notions of identifiers, data flow and namespaces, and natural language naming. We also investigate how probing results are affected by using code-specific pretraining objectives, varying the model size, or finetuning.
    Keywords Computer Science - Software Engineering ; Computer Science - Computation and Language ; Computer Science - Machine Learning
    Subject code 005
    Publishing date 2022-02-16
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  2. Book ; Online: A Simple Approach for Handling Out-of-Vocabulary Identifiers in Deep Learning for Source Code

    Chirkova, Nadezhda / Troshin, Sergey

    2020  

    Abstract: There is an emerging interest in the application of natural language processing models to source code processing tasks. One of the major problems in applying deep learning to software engineering is that source code often contains a lot of rare ... ...

    Abstract There is an emerging interest in the application of natural language processing models to source code processing tasks. One of the major problems in applying deep learning to software engineering is that source code often contains a lot of rare identifiers, resulting in huge vocabularies. We propose a simple, yet effective method, based on identifier anonymization, to handle out-of-vocabulary (OOV) identifiers. Our method can be treated as a preprocessing step and, therefore, allows for easy implementation. We show that the proposed OOV anonymization method significantly improves the performance of the Transformer in two code processing tasks: code completion and bug fixing.

    Comment: Published at the 2021 Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2021)
    Keywords Computer Science - Software Engineering ; Computer Science - Machine Learning
    Subject code 005
    Publishing date 2020-10-23
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  3. Book ; Online: Study on Precoding Optimization Algorithms in Massive MIMO System with Multi-Antenna Users

    Bobrov, Evgeny / Kropotov, Dmitry / Troshin, Sergey / Zaev, Danila

    2021  

    Abstract: The paper studies the multi-user precoding problem as a non-convex optimization problem for wireless multiple input and multiple output (MIMO) systems. In our work, we approximate the target Spectral Efficiency function with a novel computationally ... ...

    Abstract The paper studies the multi-user precoding problem as a non-convex optimization problem for wireless multiple input and multiple output (MIMO) systems. In our work, we approximate the target Spectral Efficiency function with a novel computationally simpler function. Then, we reduce the precoding problem to an unconstrained optimization task using a special differential projection method and solve it by the Quasi-Newton L-BFGS iterative procedure to achieve gains in capacity. We are testing the proposed approach in several scenarios generated using Quadriga - open-source software for generating realistic radio channel impulse response. Our method shows monotonic improvement over heuristic methods with reasonable computation time. The proposed L-BFGS optimization scheme is novel in this area and shows a significant advantage over the standard approaches. The proposed method has a simple implementation and can be a good reference for other heuristic algorithms in this field.

    Comment: 16 pages, 6 figures, 6 tables, the work has been accepted for publication in Optimization Methods and Software, comments are welcome
    Keywords Computer Science - Information Theory ; Computer Science - Networking and Internet Architecture
    Subject code 006
    Publishing date 2021-07-28
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  4. Book ; Online: Adaptive Regularized Zero-Forcing Beamforming in Massive MIMO with Multi-Antenna Users

    Bobrov, Evgeny / Chinyaev, Boris / Kuznetsov, Viktor / Lu, Hao / Minenkov, Dmitrii / Troshin, Sergey / Yudakov, Daniil / Zaev, Danila

    2021  

    Abstract: Modern wireless cellular networks use massive multiple-input multiple-output (MIMO) technology. This technology involves operations with an antenna array at a base station that simultaneously serves multiple mobile devices which also use multiple ... ...

    Abstract Modern wireless cellular networks use massive multiple-input multiple-output (MIMO) technology. This technology involves operations with an antenna array at a base station that simultaneously serves multiple mobile devices which also use multiple antennas on their side. For this, various precoding and detection techniques are used, allowing each user to receive the signal intended for him from the base station. There is an important class of linear precoding called Regularized Zero-Forcing (RZF). In this work, we propose Adaptive RZF (ARZF) with a special kind of regularization matrix with different coefficients for each layer of multi-antenna users. These regularization coefficients are defined by explicit formulas based on SVD decompositions of user channel matrices. We study the optimization problem, which is solved by the proposed algorithm, with the connection to other possible problem statements. We also compare the proposed algorithm with state-of-the-art linear precoding algorithms on simulations with the Quadriga channel model. The proposed approach provides a significant increase in quality with the same computation time as in the reference methods.

    Comment: 26 pages, 7 figures, 6 tables, prepared for the EURASIP Journal on Wireless Communications and Networking journal, comments are welcome
    Keywords Computer Science - Information Theory ; Computer Science - Networking and Internet Architecture
    Subject code 003
    Publishing date 2021-07-02
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  5. Article: [Intraoperative transesophageal electrical stimulation in cardiosurgery].

    Troshin, S V / Raskin, V V / Baialieva, A Zh / Akhundov, R N

    Anesteziologiia i reanimatologiia

    2010  , Issue 5, Page(s) 33–36

    MeSH term(s) Anesthesia, General/adverse effects ; Bradycardia/etiology ; Bradycardia/prevention & control ; Cardiac Surgical Procedures/adverse effects ; Electric Stimulation Therapy/methods ; Electrophysiologic Techniques, Cardiac/methods ; Female ; Humans ; Intraoperative Care/methods ; Male ; Middle Aged ; Treatment Outcome
    Language Russian
    Publishing date 2010-09
    Publishing country Russia (Federation)
    Document type Journal Article
    ZDB-ID 754946-5
    ISSN 0201-7563
    ISSN 0201-7563
    Database MEDical Literature Analysis and Retrieval System OnLINE

    More links

    Kategorien

  6. Book ; Online: SantaCoder

    Allal, Loubna Ben / Li, Raymond / Kocetkov, Denis / Mou, Chenghao / Akiki, Christopher / Ferrandis, Carlos Munoz / Muennighoff, Niklas / Mishra, Mayank / Gu, Alex / Dey, Manan / Umapathi, Logesh Kumar / Anderson, Carolyn Jane / Zi, Yangtian / Poirier, Joel Lamy / Schoelkopf, Hailey / Troshin, Sergey / Abulkhanov, Dmitry / Romero, Manuel / Lappert, Michael /
    De Toni, Francesco / del Río, Bernardo García / Liu, Qian / Bose, Shamik / Bhattacharyya, Urvashi / Zhuo, Terry Yue / Yu, Ian / Villegas, Paulo / Zocca, Marco / Mangrulkar, Sourab / Lansky, David / Nguyen, Huu / Contractor, Danish / Villa, Luis / Li, Jia / Bahdanau, Dzmitry / Jernite, Yacine / Hughes, Sean / Fried, Daniel / Guha, Arjun / de Vries, Harm / von Werra, Leandro

    don't reach for the stars!

    2023  

    Abstract: The BigCode project is an open-scientific collaboration working on the responsible development of large language models for code. This tech report describes the progress of the collaboration until December 2022, outlining the current state of the ... ...

    Abstract The BigCode project is an open-scientific collaboration working on the responsible development of large language models for code. This tech report describes the progress of the collaboration until December 2022, outlining the current state of the Personally Identifiable Information (PII) redaction pipeline, the experiments conducted to de-risk the model architecture, and the experiments investigating better preprocessing methods for the training data. We train 1.1B parameter models on the Java, JavaScript, and Python subsets of The Stack and evaluate them on the MultiPL-E text-to-code benchmark. We find that more aggressive filtering of near-duplicates can further boost performance and, surprisingly, that selecting files from repositories with 5+ GitHub stars deteriorates performance significantly. Our best model outperforms previous open-source multilingual code generation models (InCoder-6.7B and CodeGen-Multi-2.7B) in both left-to-right generation and infilling on the Java, JavaScript, and Python portions of MultiPL-E, despite being a substantially smaller model. All models are released under an OpenRAIL license at https://hf.co/bigcode.
    Keywords Computer Science - Software Engineering ; Computer Science - Artificial Intelligence ; Computer Science - Machine Learning
    Publishing date 2023-01-09
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  7. Book ; Online: Critical Phenomena in DIS

    Jenkovszky, L. L. / Nagy, Andrea / Troshin, S. M. / Turoci, Jolan / Tyurin, N. E.

    2010  

    Abstract: Saturation in deep inelastic scattering (DIS) and deeply virtual Compton scattering (DVCS) is associated with a phase transition between the partonic gas, typical of moderate $x$ and $Q^2$, and partonic fluid appearing at increasing $Q^2$ and decreasing ... ...

    Abstract Saturation in deep inelastic scattering (DIS) and deeply virtual Compton scattering (DVCS) is associated with a phase transition between the partonic gas, typical of moderate $x$ and $Q^2$, and partonic fluid appearing at increasing $Q^2$ and decreasing Bjorken $x$. In the statistical interpretation of DIS, the large-$x,(1-x)^n$ factor in the SF is associated with a statistical distribution (perfect gas), while the low-$x$, Regge behaved factor $x^{b(Q^2)}$ produces deviations from the perfect gas and ultimately leads to a gas-liquid phase transition. In this paper we do not intend to propose another parametrization of the structure function; instead we suggest a new insight into the internal structure of the nucleon, as seen in DIS, and its connection with that revealed in high-energy nucleons and heavy-ion collisions.

    Comment: 20 pages, 8 figures, reported at the CPOD Conference, Dubna, 2010, to be published in the International Journal of Modern Physics A
    Keywords High Energy Physics - Phenomenology ; Nuclear Theory
    Subject code 306
    Publishing date 2010-09-08
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  8. Book ; Online: Spin Phenomena in Particle Interactions

    Troshin, S.M / Tiurin, N.E / Tyurin, N.E

    1994  

    Abstract: In recent years, there has been considerable growth in research activities related to spin phenomena in high energy physics and their theoretical interpretations. It has become clear that the spin enigma is not to be considered separately but that it is ... ...

    Abstract In recent years, there has been considerable growth in research activities related to spin phenomena in high energy physics and their theoretical interpretations. It has become clear that the spin enigma is not to be considered separately but that it is strongly related to the quark-gluon structure of hadrons and their interaction dynamics.Research on spin phenomena has now attracted a significant following of experimental and theoretical physicists who meet regularly at symposiums on the topic.This book serves as an introduction to the spin puzzles at high energies. Its main focus is on spin
    Language English
    Size Online-Ressource (224 p)
    Publisher World Scientific Publishing Company
    Publishing place Singapore
    Document type Book ; Online
    Note Description based upon print version of record
    ISBN 9789810216924 ; 9810216920
    Database Library catalogue of the German National Library of Science and Technology (TIB), Hannover

    More links

    Kategorien

  9. Article: The quality of food-grade acetic acid and the consumption of potassium permanganate as a function of the processing temperature

    Provorov, A.K / Troshin, S.N

    Hydrolysis and wood chemistry USSR. 1978 (pub. 1979) (8)

    1979  

    Keywords technology
    Language English ; Russian
    Dates of publication 1979-1978
    Size p. 30-31., ill.
    Document type Article
    Database NAL-Catalogue (AGRICOLA)

    More links

    Kategorien

  10. Article: Dependence of the quality of acetic acid to be used for culinary purposes and of the consumption of potassium permanganate on the processing temperature

    Provorov, A.K / Troshin, S.N

    Gidroliznaia i lesokhimicheskaia promyshlennost'. 1978. (8)

    1978  

    Title variant Dependence of the quality of acetic acid to be used for culinary purposes and of the consumption of potassium permanganate on the processing temperature [Wood chemistry]
    Keywords wood chemistry
    Language Russian
    Size p. 18-19., ill.
    Document type Article
    Note In Russian. ; Title in original language could not be transcribed.
    ISSN 0016-9706
    Database NAL-Catalogue (AGRICOLA)

    More links

    Kategorien

To top