m

F Nous contacter


0
     
Multi angle

H 1 Algorithms in high-dimensional non-convex landscapes

Auteurs : Zdeborova, Lenka (Auteur de la Conférence)
CIRM (Editeur )

    Loading the player...

    Résumé : Analysis of algorithms in noisy high-dimensional probabilistic problems poses many current challenges. In a subclass of these problems the corresponding challenges can be overcome with the help of a method coming from statistical mechanics. I will review some of the related recent work together with progress on rigorous justification of the corresponding results.

    Keywords : machine learning # optimization # gradiant descent # statistical physics

    Codes MSC :
    62P35 - Applications to physics
    68T05 - Learning and adaptive systems
    68W25 - Approximation algorithms

      Informations sur la Vidéo

      Réalisateur : Hennenfent, Guillaume
      Langue : Anglais
      Date de publication : 26/07/2019
      Date de captation : 26/06/2019
      Collection : Research talks
      Format : MP4
      Durée : 00:56:41
      Domaine : Computer Science ; Control Theory & Optimization ; Probability & Statistics ; Geometry
      Audience : Chercheurs ; Doctorants , Post - Doctorants
      Download : https://videos.cirm-math.fr/2019-06-26_Zderborova.mp4

    Informations sur la rencontre

    Nom de la rencontre : AofA: Probabilistic, Combinatorial and Asymptotic Methods for the Analysis of Algorithms / AofA: méthodes probabilistes, combinatoires et asymptotiques pour l analyse d algorithmes
    Organisateurs de la rencontre : Bassino, Frédérique ; Martínez, Conrado ; Salvy, Bruno
    Dates : 24/06/2019 - 28/06/2019
    Année de la rencontre : 2019
    URL Congrès : https://conferences.cirm-math.fr/1940.html

    Citation Data

    DOI : 10.24350/CIRM.V.19540503
    Cite this video as: Zdeborova, Lenka (2019). Algorithms in high-dimensional non-convex landscapes. CIRM. Audiovisual resource. doi:10.24350/CIRM.V.19540503
    URI : http://dx.doi.org/10.24350/CIRM.V.19540503


    Voir aussi

    Bibliographie

    1. DIA, Mohamad, MACRIS, Nicolas, KRZAKALA, Florent, et al. Mutual information for symmetric rank-one matrix estimation: A proof of the replica formula. In : Advances in Neural Information Processing Systems. 2016. p. 424-432 - http://papers.nips.cc/paper/6379-mutual-information-for-symmetric-rank-one-matrix-estimation-a-proof-of-the-replica-formula

    2. LESIEUR, Thibault, KRZAKALA, Florent, et ZDEBOROVÁ, Lenka. Constrained low-rank matrix estimation: Phase transitions, approximate message passing and applications. Journal of Statistical Mechanics: Theory and Experiment, 2017, vol. 2017, no 7, p. 073403. - https://doi.org/10.1088/1742-5468/aa7284

    3. LESIEUR, Thibault, MIOLANE, Léo, LELARGE, Marc, et al. Statistical and computational phase transitions in spiked tensor estimation. In : 2017 IEEE International Symposium on Information Theory (ISIT). IEEE, 2017. p. 511-515. - https://doi.org/10.1109/ISIT.2017.8006580

    4. MANNELLI, Stefano Sarao, BIROLI, Giulio, CAMMAROTA, Chiara, et al. Marvels and pitfalls of the langevin algorithm in noisy high-dimensional inference. arXiv preprint arXiv:1812.09066, 2018. - https://arxiv.org/abs/1812.09066

    5. MANNELLI, Stefano Sarao, KRZAKALA, Florent, URBANI, Pierfrancesco, et al. Passed & Spurious: analysing descent algorithms and local minima in spiked matrix-tensor model. arXiv preprint arXiv:1902.00139, 2019. - https://arxiv.org/abs/1902.00139

Z