Universidade Federal de Minas Gerais

About: Universidade Federal de Minas Gerais is a education organization based out in Belo Horizonte, Minas Gerais, Brazil. It is known for research contribution in the topics: Population & Context (language use). The organization has 41631 authors who have published 75688 publications receiving 1249905 citations.

Factors associated with duration of breastfeeding.

Abstract: Objectives: To determine rates of exclusive breastfeeding and of complementary feeding and to identify variables that interfere with breastfeeding in the municipality of Itauna, MG, Brazil. Methods: A longitudinal study was undertaken enrolling 246 women who gave birth at the maternity unit of the Manoel Goncalves Hospital, in Itauna, MG. The mothers and their infants were seen monthly for the first 12 months after birth or until they stopped breastfeeding. Survival analysis procedures were used to study the duration of exclusive breastfeeding and of complementary feeding. The impact on breastfeeding duration of a series of co-variables was assessed by means of Cox regression modeling. Results: The prevalence of exclusive breastfeeding at 6 months was 5.3%, and for breastfeeding at 12 months it was33.7%.Themediandurationofexclusivebreastfeedingwas40days,andmedianbreastfeedingdurationwas237 days. Multivariate analysis demonstrated a negative association (p 9consultations),useofalcoholortobacco,delaybeforefirst feed (> 6 hours) and use of a pacifier. Conclusions: Breastfeeding rates in Itauna (MG) are well below those recommended by the World Health Organization. The principal variables with a negative relationship with duration of exclusive breastfeeding and of complementary feeding are related to mother and baby health care and, therefore, interventions are possible.

Instituições participativas e desenho institucional: algumas considerações sobre a variação da participação no Brasil democrático

Abstract: O autor analisa as instituicoes participativas (orcamentos participativos, conselhos de politicas e planos diretores municipais) surgidas na democracia brasileira recente e aponta para suas diferencas devidas a desenhos institucionais diferenciados. Estes variam na maneira como a participacao se organiza; na maneira como o Estado se relaciona com a participacao e na maneira como a legislacao exige do governo a implementacao ou nao da participacao e o autor mostra esta variacao atraves da analise destas praticas em diferentes cidades brasileiras (Porto Alegre, Sao Paulo, Belo Horizonte e Salvador). A analise dos diferentes desenhos participativos mostra que o orcamento participativo e a pratica mais democratizante, mas tambem a mais vulneravel a vontade da sociedade politica; indica ainda a importância da variacao do contexto para a efetividade da participacao.

Information-Theoretic Probing for Linguistic Structure

Abstract: The success of neural networks on a diverse set of NLP tasks has led researchers to question how much these networks actually ``know'' about natural language. Probes are a natural way of assessing this. When probing, a researcher chooses a linguistic task and trains a supervised model to predict annotations in that linguistic task from the network's learned representations. If the probe does well, the researcher may conclude that the representations encode knowledge related to the task. A commonly held belief is that using simpler models as probes is better; the logic is that simpler models will identify linguistic structure, but not learn the task itself. We propose an information-theoretic operationalization of probing as estimating mutual information that contradicts this received wisdom: one should always select the highest performing probe one can, even if it is more complex, since it will result in a tighter estimate, and thus reveal more of the linguistic information inherent in the representation. The experimental portion of our paper focuses on empirically estimating the mutual information between a linguistic property and BERT, comparing these estimates to several baselines. We evaluate on a set of ten typologically diverse languages often underrepresented in NLP research---plus English---totalling eleven languages. Our implementation is available in https://github.com/rycolab/info-theoretic-probing.