Cassava breeding and agronomy in Asia: 50 years of history and future directions.

In Asia, cassava (Manihot esculenta) is cultivated by more than 8 million farmers, driving the rural economy of many countries. The International Center for Tropical Agriculture (CIAT), in partnership with national agricultural research institutes (NARIs), instigated breeding and agronomic research in Asia, 1983. The breeding program has successfully released high-yielding cultivars resulting in an average yield increase from 13.0 t ha-1 in 1996 to 21.3 t ha-1 in 2016, with significant economic benefits. Following the success in increasing yields, cassava breeding has turned its focus to higher-value traits, such as waxy cassava, to reach new market niches. More recently, building resistance to invasive pests and diseases has become a top priority due to the emergent threat of cassava mosaic disease (CMD). The agronomic research involves driving profitability with advanced technologies focusing on better agronomic management practices thereby maintaining sustainable production systems. Remote sensing technologies are being tested for trait discovery and large-scale field evaluation of cassava. In summary, cassava breeding in Asia is driven by a combination of food and market demand with technological innovations to increase the productivity. Further, exploration in the potential of data-driven agriculture is needed to empower researchers and producers for sustainable advancement.

Early detection of reduced left ventricular systolic function by 2D speckle tracking echocardiography in patients with primary mitral regurgitation in a Vietnamese cohort.

Background: Mitral regurgitation (MR) is a common heart valve disease, causing many serious complications in several organ systems, especially the cardiovascular system. The 2D speckle tracking echocardiography (STE) is a new technique for detecting potential cardiac dysfunction when only tissue function abnormalities are present. The study aimed to assess left ventricular (LV) systolic function early by STE in patients with primary MR through global LV deformity along the global longitudinal strain (GLS). Methods: An analytical cross-sectional study was performed on 46 patients with moderate to severe primary MR as recommended by the American Society of Echocardiography (ASE) 2017. Results: The prevalence of patients with GLS reduction with ejection fraction (EF) >60%, New York Heart Association (NYHA) I, and left ventricular internal diameter systolic (LVIDs) <40 mm was 38.1%, 35.7%, and 39.5%, respectively. 100% of patients with EF<60% and LVIDs ≥40 mm had reduced GLS (<16%). The GLS index strongly correlates with the NYHA classification, degree of MR, EF, and echocardiographic parameters. Conclusion: GLS index gives a significant sign in the early detection of cardiac function abnormalities before symptoms or other echocardiographic parameters in patients with MR.

Instance-Level Few-Shot Learning With Class Hierarchy Mining.

Few-shot learning is proposed to tackle the problem of scarce training data in novel classes. However, prior works in instance-level few-shot learning have paid less attention to effectively utilizing the relationship between categories. In this paper, we exploit the hierarchical information to leverage discriminative and relevant features of base classes to effectively classify novel objects. These features are extracted from abundant data of base classes, which could be utilized to reasonably describe classes with scarce data. Specifically, we propose a novel superclass approach that automatically creates a hierarchy considering base and novel classes as fine-grained classes for few-shot instance segmentation (FSIS). Based on the hierarchical information, we design a novel framework called Soft Multiple Superclass (SMS) to extract relevant features or characteristics of classes in the same superclass. A new class assigned to the superclass is easier to classify by leveraging these relevant features. Besides, in order to effectively train the hierarchy-based-detector in FSIS, we apply the label refinement to further describe the associations between fine-grained classes. The extensive experiments demonstrate the effectiveness of our method on FSIS benchmarks. The source code is available here: https://github.com/nvakhoa/superclass-FSIS.

Influences of Material Selection, Infill Ratio, and Layer Height in the 3D Printing Cavity Process on the Surface Roughness of Printed Patterns and Casted Products in Investment Casting.

As 3D-printed (3DP) patterns are solid and durable, they can be used to create thin wall castings, which is complicated with wax patterns because of the wax's fragility and high shrinkage ratio. According to this study's experiment results, polylactic acid (PLA), polyvinyl butyral (PVB), and castable wax (CW) are suitable materials for preparing investment casting (IC) cavities. The results indicate that the casting product with the highest-quality surface is obtained using a cavity prepared using a CW-printed pattern. PLA- and PVB-printed patterns provide a good surface finish for casted products. In addition, the roughness of both the printed and casted surfaces increases as the printing layer height increases. The roughness of the casted surface varies from 2.25 µm to 29.17 µm. This investigation also considers the correlation between the infill ratio and mechanical properties of PLA-printed patterns. An increase in the infill ratios from 0% to 100% leads to a significant increase in the tensile properties of the PLA-printed pattern. The obtained results can be practically used.

Neural network structure for spatio-temporal long-term memory.

This paper proposes a neural network structure for spatio-temporal learning and recognition inspired by the long-term memory (LTM) model of the human cortex. Our structure is able to process real-valued and multidimensional sequences. This capability is attained by addressing three critical problems in sequential learning, namely the error tolerance, the significance of sequence elements and memory forgetting. We demonstrate the potential of the framework with a series of synthetic simulations and the Australian sign language (ASL) dataset. Results show that our LTM model is robust to different types of distortions. Second, our LTM model outperforms other sequential processing models in a classification task for the ASL dataset.