No abstract

No abstract

Five separate studies were conducted to determine the growth and egg production performance of Itik Pinas (IP) Kayumanggi, a new commercial hybrid egg-type mallard duck in the Philippines under an intensive management system. Results showed that the IP Kayumanggi produced by crossing male IP Khaki and female IP Itim as parents showed significant sexual dimorphism in down feather patterns. The male and female ducklings have 98% black and 99% brown down feathers upon hatch, respectively. In addition, the female ducklings have significantly higher body weight from Weeks 0–4, whereas males were heavier from Weeks 10–18. Cumulative feed consumption was 10.72 kg from brooding (Weeks 1–4) to growing (Weeks 5–16). Daily feed consumption per bird was 124 g during the pre-laying (Weeks 17–18) and laying (Weeks 19–50) periods. Phenotypic characterization revealed that males have predominantly dusky black plumage, green bill, and brown shank, whereas females have mallard-type brown plumage, black bill, and orange shank. The males have significantly higher values for all the body measurements. The average age at first lay was 18 wk, and it reached 80% egg production at 26 wk of age. Average egg weight was 67 g. The average fertility, hatchability, and hatch of fertile were 89, 55, and 62%, respectively. This study provided baseline information on the productive and reproductive performance of IP Kayumanggi.

La enseñanza de programación resulta central en los procesos de formación de los programas relacionados con las ciencias de la computación, la ingeniería de sistemas y la ingeniería de software. La formación en esta área particular del conocimiento en dichos programas tiene una particular importancia, porque de su éxito depende buena parte del desempeño académico de los estudiantes en asignaturas futuras del programa. Sin embargo, durante los últimos 20 años, la mayor parte de los procesos educativos han mantenido el mismo enfoque con resultados mezclados. En este trabajo se presenta una breve reseña de los enfoques tanto conceptuales como metodológicos aplicados históricamente, y se contextualiza la situación presente en el Politécnico Grancolombiano en torno a la formación en programación con énfasis en las dificultades actuales; y se presenta una alternativa de aplicación de TIC bajo la forma de una herramienta de software para la enseñanza de conceptos básicos de programación que puede facilitar los procesos de enseñanza – aprendizaje en las asignaturas asociadas.

Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization of the loss function to make the effect of most parameters negligible. In this work, we propose instead to use regularization for preventing neurons from dying or becoming linear, a technique which we denote as jumpstart regularization. In comparison to conventional training, we obtain neural networks that are thinner, deeper , and-most importantly-more parameter-efficient.