Advances in biological and medical technologies have been providing us explosive volumes of biological and physiological data, such as medical images, electroencephalography, genomic and protein sequences. Learning from these data facilitates the understanding of human health and disease. Developed from artificial neural networks, deep learning-based algorithms show great promise in extracting features and learning patterns from complex data. The aim of this paper is to provide an overview of deep learning techniques and some of the state-of-the-art applications in the biomedical field. We first introduce the development of artificial neural network and deep learning. We then describe two main components of deep learning, i.e., deep learning architectures and model optimization. Subsequently, some examples are demonstrated for deep learning applications, including medical image classification, genomic sequence analysis, as well as protein structure classification and prediction. Finally, we offer our perspectives for the future directions in the field of deep learning.

Achieving good speed and accuracy trade-off on a target platform is very important in deploying deep neural networks in real world scenarios. However, most existing automatic architecture search approaches only concentrate on high performance. In this work, we propose an algorithm that can offer better speed/accuracy trade-off of searched networks, which is termed "Partial Order Pruning". It prunes the architecture search space with a partial order assumption to automatically search for the architectures with the best speed and accuracy trade-off. Our algorithm explicitly takes profile information about the inference speed on the target platform into consideration. With the proposed algorithm, we present several Dongfeng (DF) networks that provide high accuracy and fast inference speed on various application GPU platforms. By further searching decoder architectures, our DF-Seg real-time segmentation networks yield state-of-the-art speed/accuracy trade-off on both the target embedded device and the high-end GPU.

No abstract

No abstract

Plant invasion is one of the major threats to natural ecosystems. Phenotypic plasticity is considered to be important for promoting plant invasiveness. High tolerance of stress can also increase survival of invasive plants in adverse habitats. Limited growth and conservation of carbohydrate are considered to increase tolerance of flooding in plants. However, few studies have examined whether invasive species shows a higher phenotypic plasticity in response to waterlogging or a higher tolerance of waterlogging (lower plasticity) than native species. We conducted a greenhouse experiment to compare the growth and morphological and physiological responses to waterlogging of the invasive, clonal, wetland species Alternanthera philoxeroides with those of its co-occurring, native, congeneric, clonal species Alternanthera sessilis. Plants of A. philoxeroides and A. sessilis were subjected to three treatments (control, 0 and 60 cm waterlogging). Both A. philoxeroides and A. sessilis survived all treatments. Overall growth was lower in A. philoxeroides than in A. sessilis, but waterlogging negatively affected the growth of A. philoxeroides less strongly than that of A. sessilis. Alternanthera philoxeroides thus showed less sensitivity of growth traits (lower plasticity) and higher waterlogging tolerance. Moreover, the photosynthetic capacity of A. philoxeroides was higher than that of A. sessilis during waterlogging. Alternanthera philoxeroides also had higher total non-structural and non-soluble carbohydrate concentrations than A. sessilis at the end of treatments. Our results suggest that higher tolerance to waterlogging and higher photosynthetic capacity may partly explain the invasion success of A. philoxeroides in wetlands.