Navegando por Palavras-chave "Hybrid"

Agora exibindo 1 - 2 de 2
  • Nenhuma Miniatura disponível
    Item
    Somente Metadadados
    Automatic feature engineering for regression models with machine learning: An evolutionary computation and statistics hybrid
    (Elsevier Science Inc, 2018) de Melo, Vinicius Veloso [UNIFESP]; Banzhaf, Wolfgang
    Symbolic Regression (SR) is a well-studied task in Evolutionary Computation (EC), where adequate free-form mathematical models must be automatically discovered from observed data. Statisticians, engineers, and general data scientists still prefer traditional regression methods over EC methods because of the solid mathematical foundations, the interpretability of the models, and the lack of randomness, even though such deterministic methods tend to provide lower quality prediction than stochastic EC methods. On the other hand, while EC solutions can be big and uninterpretable, they can be created with less bias, finding high-quality solutions that would be avoided by human researchers. Another interesting possibility is using EC methods to perform automatic feature engineering for a deterministic regression method instead of evolving a single model
  • Nenhuma Miniatura disponível
    Item
    Somente Metadadados
    Pressure-driven opening of carbon nanotubes
    (Royal soc chemistry, 2016) Chaban, Vitaly V. [UNIFESP]; Prezhdo, Oleg V.
    The closing and opening of carbon nanotubes (CNTs) is essential for their applications in nanoscale chemistry and biology. We report reactive molecular dynamics simulations of CNT opening triggered by internal pressure of encapsulated gas molecules. Confined argon generates 4000 bars of pressure inside capped CNT and lowers the opening temperature by 200 K. Chemical interactions greatly enhance the efficiency of CNT opening: fluorine-filled CNTs open by fluorination of carbon bonds at temperature and pressure that are 700 K and 1000 bar lower than for argon-filled CNTs. Moreover, pressure induced CNT opening by confined gases leaves the CNT cylinders intact and removes only the fullerene caps, while the empty CNT decomposes completely. In practice, the increase in pressure can be achieved by near-infrared light, which penetrates through water and biological tissues and is absorbed by CNTs, resulting in rapid local heating. Spanning over a thousand of bars and Kelvin, the reactive and non-reactive scenarios of CNT opening represent extreme cases and allow for a broad experimental control over properties of the CNT interior and release conditions of the confined species. The detailed insights into the thermodynamic conditions and chemical mechanisms of the pressure-induced CNT opening provide practical guidelines for the development of novel nanoreactors, catalysts, photo-catalysts, imaging labels and drug delivery vehicles.