PAC-Bayesian generalization bounds are among the strongest distribution-dependent guarantees in statistical learning and are typically expressed in terms of the Kullback-Leibler (KL) divergence between a prior and posterior distribution. While analytically convenient, the KL divergence requires the posterior to be absolutely continuous w.r.t. the prior, imposing support constraints that can substantially restrict the choice of posterior. Moreover, the KL divergence does not account for the underlying geometry of the probability space. In this work, we derive new PAC-Bayesian bounds replacing the KL divergence with general optimal transport (OT) distances, including the total variation and Wasserstein-p distances.
In this paper, a new three-parameter lifetime distribution is derived from the log-cosinepower-generated family of distributions with the modified Lindley distribution as the baseline. It aims to modify the functionalities of the modified Lindley distribution by using a logarithmic-cosine transformation. Its key mathematical properties are established, and parameter estimation is performed using both maximum likelihood and Bayesian approaches under a variety of loss functions. A simulation study evaluates the performance of the classical and Bayesian estimators using mean squared errors and posterior risks, respectively, supplemented by a diagnostic study with three parallel chains to ensure Markov chain Monte Carlo convergence. In addition, three real data sets are analyzed to demonstrate the flexibility and practical applicability of the new distribution, featuring comprehensive Bayesian inference and prior sensitivity analysis.
OBJECTIVE:Symphyseal disjunction is a rare event in childbirth, occurring almost exclusively after vaginal delivery. Radiologically, it is defined by a symphyseal gap greater than 10 mm. No comparative studies have been done on the potential risk factors involved in its development. The aim of the present study was to describe the circumstances of occurrence and risk factors for symphyseal disjunction. METHODS:This was an exploratory case-control study (1:4). Cases were from literature since January 2010 to December 2023 were joined with disjunctions observed in our center between April 2021 and April 2023. Controls were randomly selected from vaginal deliveries in our center during the same period. RESULTS:A total of 20 cases were observed (12 in the literature, 8 in our center). Pain was constant and gait disturbance frequent (85%). Disjunctions were mostly diagnosed a few days after birth (median 3.5 [1; 5.5] [extremes 0.5-90 days]). Surgical osteosynthesis was required in one quarter of the cases. Comparison with controls showed that neither parity, neonatal weight, instrumental delivery nor delivery conditions were associated with the occurrence of these disjunctions. On the other hand, neonatal head circumference at birth was significantly higher in the disjunction group (35 [34; 36] vs. 33 [32; 34] cm, P < 0.001). Significantly higher values of biparietal diameter were also observed in cases of disjunction compared to controls (84.1 [82.0; 86.1] vs. 81.4 [80.0; 82.9] mm, P = 0.005). CONCLUSION:Symphyseal disjunction is a rare event most often presenting with pain and difficulty in walking after vaginal delivery. Its occurrence seems to be related to the fetal head circumference, and not to the conditions of vaginal delivery.
In empirical studies, data are often truncated because of institutional limits, measurement problems, or sampling methods, which means observations are limited to a certain range. If truncation isn't considered, parameter estimates can be skewed, and statistical conclusions can be wrong. Because of this, truncated probability distributions are needed to accurately model real-world data. This paper gives a bibliometric review of studies on truncated distributions to show the field's history, structure, and how its topics have changed. Data from Scopus between 2000 and 2025 was gathered and examined using the Bibliometrix package in R. Publication trends, key authors and sources, cooperation, and new subjects were assessed using descriptive stats and science-mapping methods like co-authorship, co-citation, keyword co-occurrence, topic mapping, and historical analysis. The results show consistent growth in scientific output, high citation rates for key theoretical work, and more interdisciplinary use in fields like economics, medical statistics, survival analysis, reliability engineering, and finance. The results show a research area that is both well-established and still changing, with solid theoretical bases, more practical uses, and more global teamwork. This work gives a structured look at research on truncated distributions and provides helpful information for people who want to find key studies, main subjects, and future research directions in the study of truncated data.
This article introduces new one-parameter discrete trigonometric distributions for analyzing count data. In particular, it focuses on a simple, one-parameter discrete trigonometric version of the Lindley distribution, known as the discrete Sin-Lindley distribution. The key mathematical properties are derived, including the probability mass function, cumulative distribution function, quantile function, probability generating function, moments, skewness, kurtosis, and order statistics. The maximum likelihood approach is then employed to estimate the unique parameter. Simulation studies demonstrate the effectiveness of the new model across varying sample sizes. The applicability and robustness of the model are demonstrated by analyzing five real-world datasets and by comparing it with Lindley-related distributions and discrete trigonometric distributions, highlighting its potential in statistical analysis.