Buenos Aires, Provincia de Buenos Aires, Argentina
436 seguidores 422 os

Unirse para ver el perfil

Acerca de

I'm a Computer Science graduate currently working as Software Engineer / Data Engineer…

Actividad

Unirse para ver toda la actividad

Experiencia y educación

  • SecurityScorecard

Mira la experiencia completa de Fabrizio Sebastian

Mira su cargo, antigüedad y más

o

Al hacer clic en «Continuar» para unirte o iniciar sesión, aceptas las Condiciones de uso, la Política de privacidad y la Política de cookies de LinkedIn.

Licencias y certificaciones

Únete para ver todas las certificaciones

Proyectos

  • DeltaML

    -

    DeltaML is a decentralized platform for machine learning that preserves the privacy of the s.
    It uses Federated Learning as scheme for training models without needing to see the actual data of the s involved, Homomorphic Encryption for securing the model during the training (the trainers can't steal it because all the operations are done over the encrypted model), and Blockchain + Smart Contracts to distribute payments between the participants involved according to how much they…

    DeltaML is a decentralized platform for machine learning that preserves the privacy of the s.
    It uses Federated Learning as scheme for training models without needing to see the actual data of the s involved, Homomorphic Encryption for securing the model during the training (the trainers can't steal it because all the operations are done over the encrypted model), and Blockchain + Smart Contracts to distribute payments between the participants involved according to how much they contributed to the process (without the need of having to trust to a third party to enforce the payment logic).

    Otros creadores
    Ver proyecto

  • Text classification algorithm

    -

    Design and development of a scalable algorithm with the capability of classify new text documents in categories generated by applying clustering techniques (non-supervised learning).

    Representing the documents as vectors where each component is a term, and the value stored in that component is the metric TFxIDF that weights the term using the global weight (rarity in the whole text corpus) and the local weight (term frequency in a document).

    Preprocessing of the data (reduction of…

    Design and development of a scalable algorithm with the capability of classify new text documents in categories generated by applying clustering techniques (non-supervised learning).

    Representing the documents as vectors where each component is a term, and the value stored in that component is the metric TFxIDF that weights the term using the global weight (rarity in the whole text corpus) and the local weight (term frequency in a document).

    Preprocessing of the data (reduction of the dimentionality) by removing stopwords, applying the Porter algorithm for stemming, removing the components with lower weight (TFxIDF).

    Implemented a two-step algorithm. First step, choosing a random sample of the documents as seeds with the size of the root square of the total amount of points times the amount of clusters needed. using a hierarchical agglomerative clustering algorithm. Second step, K-Means algorithm using the seeds chosen in the previous step.

    the cosine distance for calculating similarity between documents.

    Team project for the course Data Organization ("Organizacion de Datos"). The course covers themes related to machine learning, cryptography, data compression and scalability of algorithms.

    Otros creadores
    Ver proyecto

Idiomas

  • Spanish

    Competencia bilingüe o nativa

  • English

    Competencia profesional completa

Más actividad de Fabrizio Sebastian

Ver el perfil completo de Fabrizio Sebastian

  • Descubrir a quién conocéis en común
  • Conseguir una presentación
  • ar con Fabrizio Sebastian directamente
Unirse para ver el perfil completo

Perfiles similares

Añade nuevas aptitudes con estos cursos

Ver todos los cursos