Centro Universitario Didaskalos

Transformer models have greatly changed how deep learning works, particularly in natural language processing, computer vision, and multi-modal learning problems. While these models have done great empirically, the mathematical foundations and stability of these models remain poorly understood. In this paper, we perform a holistic analysis of the mathematical foundations of the Transformer architecture such as self-attention and positional encodings, and mathematical properties of layer normalization. Specifically, our work explores the stability conditions given different settings of training and inference, using principles from linear algebra, spectral analysis, and dynamical systems. In our analysis, we elaborate on how architectural parameters (e.g., depth, width, and initialization scheme) could influence the behaviour of gradients, convergence, and robustness. We also investigate the conditions by which we can assure stable propagation of information through layers and identified the conditions in which transformers exhibit vanishing or exploding gradients. We hope our work will offer reliable theoretical prospects of robustness for practical implementations of Transformer-based systems, and offer ways to practically choose a model that is stable and demonstrably efficient. The paper offers a connection for theoretical analysis and practical implementation, which are the foundations we believe will facilitate the next generation of theoretically-based grounded deep learning architectures.

The activity of binary pattern-based texture descriptors for hand posture recognition is limited by factors such as complex background, illumination, rotation and scale invariant. In this aspect, high potential texture descriptor is necessary for robust hand posture recognition. The proposed work, local geometric discriminative directional pattern (LGDDP) is based on geometric discriminative features such as mean curvature and local directional pattern (LDP) for efficient recognition system. The geometric structure characteristic of hand signs images through mean curvature is estimated by derivative filters. Then, the result of local geometric discriminant (GD) is turned into binary response to integrate edge details of LDP in an image. The experimental results of these texture descriptors are validated by Extreme Learning Machine (ELM) classifier using the datasets such as Jochen Triesch of uniform, dark, complex background, and National University of Singapore (NUS-I). The proposed work shows the pertinent feature representation for efficient hand sign recognition against other local descriptors and achieves the higher classification rate.

Abstract INTRODUCTION. Actinomycosis is a rare infection, caused by Gram positive anaerobic and filamentous bacteria that usually colonize human mucosa. Infection is produced chronically in immunosuppressed or immunocompetent patients. We report the case of a 41 year old male presenting chronic cough, diagnosed with Atypical Pneumonia with histopathological compatibility with infection by Actinomyces Sp. CLINICAL CASE. A 41-year-old male from Toluca, Estado de México, attended an evaluation after 2 months of chronic cough, associated with dyspnea and yellow sputum. Known with gastroesophageal reflux disease, rhinitis and asthma on partial control with inhaled and systemic corticoids and bronchodilators. Before referral, he received empiric antibiotics with no response. Helicoidal CT was performed with multiple nodules, patchy condensates in the lower right lobe pulmonary and right hilar calcifications. A broncho alveolar lavage and transbronchial lung biopsies were performed which revealed respiratory mucosa with severe chronic and acute inflammation associated with infection caused by organisms that are morphologically compatible with Actinomyces Sp. The patient receives treatment for 4 weeks with intravenous ceftriaxone and was given to the patient subsequently treatment with oral amoxicillin for 6 month with resolution of the Pneumonia clinically and by image. The patient has been followed up with pulmonary function tests which improved with time and no recurrence of the infection was noted. The patient is currently asymptomatic, with no evidence of systemic inflammatory response and with good control of rhinitis and asthma.CONCLUSION: Infection by Actinomyces Sp. is rare, usually presenting as a chronic suppurative infection. Immunosuppression is a classic risk factor, but infection can happen in the immunocompetent. The imaging for actinomycosis is nonspecific. In the acute presentation, it can look like any other pneumonia. In chronic forms, it can present as a pulmonary mass or can cavitate. Epidemiology hasn't been established because of the low incidence of this disease.

The nipple-areolar complex (NAC) is important from an aesthetic and functional standpoint for both sexes; however, methodology for NAC placement in men is less established. The objectives of this research are to determine normal measurements for NAC in the male Mexican population, its association with body mass index (BMI), and age, as well as presenting a reliable linear regression model for preoperative planning and placement. This study was carried out on the male population of the State of Mexico. NAC anthropometry was performed on 385 adult males. Descriptive statistics were used to determine measures of central tendency and dispersion. Spearman correlation test was performed in order to evaluate the linear association between age and BMI with anthropometry, and a linear regression model was generated for the measured variables. Finally, the predicted and measured values were compared. BMI is strongly correlated with NAC anthropometry, an association that is independent from age. NAC diameter was also independent from BMI and age, 2.25cm being the average. We provide the anthropometric values obtained, and a linear regression equationwhich might be used in preoperative planning to determine optimal NAC placement. This method covers a wide range of BMI and age values. This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

Economic reforms have altered Vietnam from one of the world's lowest income countries to a lower-middle-income country. However, since 2008, Vietnam has faced major challenges in restoring macroeconomic balances and sustaining high economic growth. Global integration has led to greater economic volatility, and Vietnam has become more susceptible to output fluctuations among its trading partners at a time when the latter are experiencing significant recession. Vietnamese authorities have responded by adopting a stimulus package, which has temporarily supported growth, but done little to restore long-term macro-stability. Light manufacturing has emerged as an important source of growth and productive employment in economies with a comparative advantage in labor-intensive sectors, as is the case in Vietnam, and thus presents great potential for improving the country's economy. This book focuses on five light manufacturing industries that form the core of the Vietnamese industrial sector: processed foods and beverages, leather, wood processing and wood products, metal products, and apparel. It aims to examine the following issues: 1) the binding constraints facing light manufacturing in Vietnam; 2) the coping mechanisms of firms to these constraints; and 3) the practical policy reforms which would help firms overcome the constraints and jump-start industry to facilitate the transformation of Vietnam into a modern economy.