LIVIVO - The Search Portal for Life Sciences

zur deutschen Oberfläche wechseln
Advanced search

Search results

Result 1 - 6 of total 6

Search options

  1. Book ; Online: Revisiting Data Augmentation for Rotational Invariance in Convolutional Neural Networks

    Quiroga, Facundo Manuel / Ronchetti, Franco / Lanzarini, Laura / Fernandez-Bariviera, Aurelio

    2023  

    Abstract: Convolutional Neural Networks (CNN) offer state of the art performance in various computer vision tasks. Many of those tasks require different subtypes of affine invariances (scale, rotational, translational) to image transformations. Convolutional ... ...

    Abstract Convolutional Neural Networks (CNN) offer state of the art performance in various computer vision tasks. Many of those tasks require different subtypes of affine invariances (scale, rotational, translational) to image transformations. Convolutional layers are translation equivariant by design, but in their basic form lack invariances. In this work we investigate how best to include rotational invariance in a CNN for image classification. Our experiments show that networks trained with data augmentation alone can classify rotated images nearly as well as in the normal unrotated case; this increase in representational power comes only at the cost of training time. We also compare data augmentation versus two modified CNN models for achieving rotational invariance or equivariance, Spatial Transformer Networks and Group Equivariant CNNs, finding no significant accuracy increase with these specialized methods. In the case of data augmented networks, we also analyze which layers help the network to encode the rotational invariance, which is important for understanding its limitations and how to best retrain a network with data augmentation to achieve invariance to rotation.
    Keywords Computer Science - Computer Vision and Pattern Recognition ; Computer Science - Neural and Evolutionary Computing ; I.2.10 ; I.5.2
    Subject code 006
    Publishing date 2023-10-12
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  2. Article: The inefficiency of Bitcoin revisited

    Fernández Bariviera, Aurelio

    Economics letters Vol. 161 , p. 1-4

    a dynamic approach

    2017  Volume 161, Page(s) 1–4

    Author's details Aurelio F. Bariviera
    Keywords Bitcoin ; Long range dependence ; Volatility ; Hurst exponent
    Language English
    Publisher Elsevier
    Publishing place Amsterdam [u.a.]
    Document type Article
    ZDB-ID 717210-2
    ISSN 0165-1765
    Database ECONomics Information System

    More links

    Kategorien

  3. Article: Informational efficiency in distressed markets

    Fernández Bariviera, Aurelio Fernández / Guercio, M. Belén / Martinez, Lisana B

    The economic and social review Vol. 45, No. 3 , p. 349-369

    the case of European corporate bonds

    2014  Volume 45, Issue 3, Page(s) 349–369

    Author's details Aurelio Fernández Bariviera; M. Belén Guercio; Lisana B. Martinez
    Keywords Unternehmensanleihe ; Finanzkrise ; Effizienzmarkthypothese ; Europa
    Language English
    Size graph. Darst.
    Publishing place Dublin
    Document type Article
    ZDB-ID 862764-2 ; 2076985-4
    ISSN 0012-9984
    ISSN 0012-9984
    Database ECONomics Information System

    More links

    Kategorien

  4. Article ; Online: Thermodynamics of firms' growth.

    Zambrano, Eduardo / Hernando, Alberto / Fernández Bariviera, Aurelio / Hernando, Ricardo / Plastino, Angelo

    Journal of the Royal Society, Interface

    2015  Volume 12, Issue 112

    Abstract: The distribution of firms' growth and firms' sizes is a topic under intense scrutiny. In this paper, we show that a thermodynamic model based on the maximum entropy principle, with dynamical prior information, can be constructed that adequately describes ...

    Abstract The distribution of firms' growth and firms' sizes is a topic under intense scrutiny. In this paper, we show that a thermodynamic model based on the maximum entropy principle, with dynamical prior information, can be constructed that adequately describes the dynamics and distribution of firms' growth. Our theoretical framework is tested against a comprehensive database of Spanish firms, which covers, to a very large extent, Spain's economic activity, with a total of 1,155,142 firms evolving along a full decade. We show that the empirical exponent of Pareto's law, a rule often observed in the rank distribution of large-size firms, is explained by the capacity of economic system for creating/destroying firms, and that can be used to measure the health of a capitalist-based economy. Indeed, our model predicts that when the exponent is larger than 1, creation of firms is favoured; when it is smaller than 1, destruction of firms is favoured instead; and when it equals 1 (matching Zipf's law), the system is in a full macroeconomic equilibrium, entailing 'free' creation and/or destruction of firms. For medium and smaller firm sizes, the dynamical regime changes, the whole distribution can no longer be fitted to a single simple analytical form and numerical prediction is required. Our model constitutes the basis for a full predictive framework regarding the economic evolution of an ensemble of firms. Such a structure can be potentially used to develop simulations and test hypothetical scenarios, such as economic crisis or the response to specific policy measures.
    MeSH term(s) Commerce ; Humans ; Models, Economic ; Spain
    Language English
    Publishing date 2015-10-28
    Publishing country England
    Document type Journal Article
    ZDB-ID 2156283-0
    ISSN 1742-5662 ; 1742-5689
    ISSN (online) 1742-5662
    ISSN 1742-5689
    DOI 10.1098/rsif.2015.0789
    Database MEDical Literature Analysis and Retrieval System OnLINE

    More links

    Kategorien

  5. Book ; Online: Thermodynamics of firms' growth

    Zambrano, Eduardo / Hernando, Alberto / Fernandez-Bariviera, Aurelio / Hernando, Ricardo / Plastino, Angelo

    2015  

    Abstract: The distribution of firms' growth and firms' sizes is a topic under intense scrutiny. In this paper we show that a thermodynamic model based on the Maximum Entropy Principle, with dynamical prior information, can be constructed that adequately describes ... ...

    Abstract The distribution of firms' growth and firms' sizes is a topic under intense scrutiny. In this paper we show that a thermodynamic model based on the Maximum Entropy Principle, with dynamical prior information, can be constructed that adequately describes the dynamics and distribution of firms' growth. Our theoretical framework is tested against a comprehensive data-base of Spanish firms, which covers to a very large extent Spain's economic activity with a total of 1,155,142 firms evolving along a full decade. We show that the empirical exponent of Pareto's law, a rule often observed in the rank distribution of large-size firms, is explained by the capacity of the economic system for creating/destroying firms, and can be used to measure the health of a capitalist-based economy. Indeed, our model predicts that when the exponent is larger that 1, creation of firms is favored; when it is smaller that 1, destruction of firms is favored instead; and when it equals 1 (matching Zipf's law), the system is in a full macroeconomic equilibrium, entailing "free" creation and/or destruction of firms. For medium and smaller firm-sizes, the dynamical regime changes; the whole distribution can no longer be fitted to a single simple analytic form and numerical prediction is required. Our model constitutes the basis of a full predictive framework for the economic evolution of an ensemble of firms that can be potentially used to develop simulations and test hypothetical scenarios, as economic crisis or the response to specific policy measures.
    Keywords Nonlinear Sciences - Adaptation and Self-Organizing Systems ; Physics - Physics and Society ; Quantitative Finance - General Finance
    Subject code 338
    Publishing date 2015-04-28
    Publishing country us
    Document type Book ; Online
    Database BASE - Bielefeld Academic Search Engine (life sciences selection)

    More links

    Kategorien

  6. Article: Immunization strategy in a fuzzy environment

    Terceño Gómez, Antonio / Brotons Martínez, José Manuel / Fernández Bariviera, Aurelio

    Fuzzy economic review : the review of the International Association for Fuzzy-Set Management and Economy Vol. 12, No. 2 , p. 95-116

    2007  Volume 12, Issue 2, Page(s) 95–116

    Author's details A. Terceño Gómez; J. M. Brotons Martínez; A. Fernández Bariviera
    Language English
    Size graph. Darst.
    Publishing place Reus
    Document type Article
    ZDB-ID 2097221-0
    Database ECONomics Information System

    More links

    Kategorien

To top