Number of the records: 1  

On the Non-Learnability of a Single Spiking Neuron

  1. 1.
    SYSNO ASEP0405292
    Document TypeJ - Journal Article
    R&D Document TypeJournal Article
    Subsidiary JČlánek ve WOS
    TitleOn the Non-Learnability of a Single Spiking Neuron
    TitleO nenaučitelnosti jednoho spiking neuronu
    Author(s) Šíma, Jiří (UIVT-O) RID, SAI, ORCID
    Sgall, Jiří (MU-W) RID, ORCID, SAI
    Source TitleNeural Computation - ISSN 0899-7667
    Roč. 17, č. 12 (2005), s. 2635-2647
    Number of pages13 s.
    Languageeng - English
    CountryUS - United States
    Keywordsspiking neuron ; consistency problem ; NP-completness ; PAC model ; robust learning ; representation problem
    Subject RIVBA - General Mathematics
    R&D ProjectsGA201/02/1456 GA ČR - Czech Science Foundation (CSF)
    1ET100300517 GA AV ČR - Academy of Sciences of the Czech Republic (AV ČR)
    LN00A056 GA MŠMT - Ministry of Education, Youth and Sports (MEYS)
    1M0545 GA MŠMT - Ministry of Education, Youth and Sports (MEYS)
    CEZAV0Z10300504 - UIVT-O (2005-2011)
    AV0Z10190503 - MU-W (2005-2011)
    UT WOS000232530300005
    EID SCOPUS27144501832
    DOI10.1162/089976605774320601
    AnnotationThe computational complexity of training a single spiking neuron N with adjustable synaptic delays and binary coded inputs and output is studied. A synchronization technique is introduced so that the results concerning the non-learnability of spiking neurons with binary delays are generalized to arbitrary delays. In particular, the consistency problem for N with programmable delays and its approximation version are proven to be NP-complete. It follows that the spiking neurons with arbitrary synaptic delays are not properly PAC-learnable and do not allow robust learning unless RP=NP. In addition, the representation problem for N, i.e., a question whether an n-variable Boolean function given in DNF (or as a disjunction of O(n) threshold gates) can be computed by a spiking neuron, is shown to be coNP-hard.
    WorkplaceInstitute of Computer Science
    ContactTereza Šírová, sirova@cs.cas.cz, Tel.: 266 053 800
    Year of Publishing2006

Number of the records: 1  

  This site uses cookies to make them easier to browse. Learn more about how we use cookies.