Brandière, Odile; Duflo, Marie Do stochastic algorithms circumvent traps? (Les algorithmes stochastiques contournent-ils les pièges?) (French) Zbl 0841.68048 C. R. Acad. Sci., Paris, Sér. I 321, No. 3, 335-338 (1995). Summary: We are considering the validity of commonly held claims: “a stochastic gradient algorithm only converges towards one of local minima” or “a stochastic algorithm only converges towards an asymptotically stable solution of the associated differential equation”. We prove that a stochastic algorithm does not fall into a regular trap if the noise is exciting in a repulsive direction. Cited in 1 Document MSC: 68W10 Parallel algorithms in computer science Keywords:stochastic gradient algorithm PDFBibTeX XMLCite \textit{O. Brandière} and \textit{M. Duflo}, C. R. Acad. Sci., Paris, Sér. I 321, No. 3, 335--338 (1995; Zbl 0841.68048)