Showing papers by "Viviana Consonni published in 2021"
Research Triangle Park1, Oak Ridge Institute for Science and Education2, University of Bari3, L'Oréal4, University of Cambridge5, University of North Carolina at Chapel Hill6, Universidade Federal de Goiás7, Dow Chemical Company8, University of Milano-Bicocca9, Mario Negri Institute for Pharmacological Research10, Michigan State University11, Kyoto University12, Henry M. Jackson Foundation for the Advancement of Military Medicine13, Rutgers University14, Simulations Plus, Inc.15, Skolkovo Institute of Science and Technology16, North Carolina State University17, Pacific Northwest National Laboratory18, UL19, University of Strasbourg20, United States Environmental Protection Agency21, National Institutes of Health22, U.S. Agency for Toxic Substances and Disease Registry23, East China University of Science and Technology24, University of Colorado Boulder25, Dalian University of Technology26, Dow AgroSciences27
TL;DR: In this paper, the authors proposed a method to assess tens of thousands of chemical substances that need to be assessed for their potential toxicity, which serves as the basis for regulatory testing.
Abstract: Background: Humans are exposed to tens of thousands of chemical substances that need to be assessed for their potential toxicity. Acute systemic toxicity testing serves as the basis for regulatory ...
38 citations
TL;DR: The regression toolbox for MATLAB is described, which is a collection of modules for calculating some well-known regression methods: Ordinary Least Squares, Partial Le least Squares (PLS), Principal Component Regression (PCR), Ridge and local regression based on sample similarities, such as Binned Nearest Neighbours and k-NearestNeighbours regression methods.
14 citations
TL;DR: DNA mini-barcoding and infrared spectroscopy failed in the taxonomic distinction, whereas enabled PLS-DA classification of aged samples (and thus the identification of fresh prepared artisanal products) with model non-error rate in fitting equal to 0.96 and 0.89.
Abstract: Tuna “Bottarga” is an artisanal food speciality prepared from the roe of Bluefin Tuna (Thunnus thynnus). Due to the recent restriction of Tuna catches fixed by EU, cheaper substitutes from yellowfin tuna (Thunnus albacares) are currently available in the mass distribution. Since the product is sold sliced or grated, morphological distinctive features may be lost, and fraudulent substitution is possible. In this study we tested DNA mini-barcoding and infrared spectroscopy as tools to trace its authenticity. A certified chemical standard, fresh (n = 4) and processed (n = 11) ovarian identified by the morphological observation of the tuna catches, commercial bottargasamples (n = 5) and artificial aged aliquots of all these samples (n = 50) were used in the experiments. The results showed that infrared spectroscopy failed in the taxonomic distinction, whereas enabled PLS-DA classification of aged samples (and thus the identification of fresh prepared artisanal products) with model non-error rate in fitting equal to 0.96, and 0.89 in cross-validation. Conversely, DNA mini-barcoding with mtDNA control region, success in species identification (100% maximum identity by the standard comparison BLAST approach against GenBank), regardless the processing state of the product.
8 citations
TL;DR: In this paper, the authors compared four of the most widely used approaches for tuning hyperparameters, namely, grid search, random search, tree-structured Parzen estimator, and genetic algorithms on three multitask QSAR datasets.
Abstract: Neural networks are rapidly gaining popularity in chemical modeling and Quantitative Structure–Activity Relationship (QSAR) thanks to their ability to handle multitask problems. However, outcomes of neural networks depend on the tuning of several hyperparameters, whose small variations can often strongly affect their performance. Hence, optimization is a fundamental step in training neural networks although, in many cases, it can be very expensive from a computational point of view. In this study, we compared four of the most widely used approaches for tuning hyperparameters, namely, grid search, random search, tree-structured Parzen estimator, and genetic algorithms on three multitask QSAR datasets. We mainly focused on parsimonious optimization and thus not only on the performance of neural networks, but also the computational time that was taken into account. Furthermore, since the optimization approaches do not directly provide information about the influence of hyperparameters, we applied experimental design strategies to determine their effects on the neural network performance. We found that genetic algorithms, tree-structured Parzen estimator, and random search require on average 0.08% of the hours required by grid search; in addition, tree-structured Parzen estimator and genetic algorithms provide better results than random search.
5 citations