Michael A. Chapman

In this paper, we designed an end-to-end spectral-spatial residual network (SSRN) that takes raw 3-D cubes as input data without feature engineering for hyperspectral image classification. In this network, the spectral and spatial residual blocks consecutively learn discriminative features from abundant spectral signatures and spatial contexts in hyperspectral imagery (HSI). The proposed SSRN is a supervised deep learning framework that alleviates the declining-accuracy phenomenon of other deep learning models. Specifically, the residual blocks connect every other 3-D convolutional layer through identity mapping, which facilitates the backpropagation of gradients. Furthermore, we impose batch normalization on every convolutional layer to regularize the learning process and improve the classification performance of trained models. Quantitative and qualitative results demonstrate that the SSRN achieved the state-of-the-art HSI classification accuracy in agricultural, rural-urban, and urban data sets: Indian Pines, Kennedy Space Center, and University of Pavia.

Multiple myeloma is an incurable malignancy of plasma cells, and its pathogenesis is poorly understood. Here we report the massively parallel sequencing of 38 tumour genomes and their comparison to matched normal DNAs. Several new and unexpected oncogenic mechanisms were suggested by the pattern of somatic mutation across the data set. These include the mutation of genes involved in protein translation (seen in nearly half of the patients), genes involved in histone methylation, and genes involved in blood coagulation. In addition, a broader than anticipated role of NF-κB signalling was indicated by mutations in 11 members of the NF-κB pathway. Of potential immediate clinical relevance, activating mutations of the kinase BRAF were observed in 4% of patients, suggesting the evaluation of BRAF inhibitors in multiple myeloma clinical trials. These results indicate that cancer genome sequencing of large collections of samples will yield new insights into cancer not anticipated by existing knowledge. Multiple myeloma, a malignancy of plasma cells, remains incurable and is poorly understood. Chapman et al. have used next-generation sequencing to compare 38 multiple myeloma genomes with those of normal cells from the same patients. The disease involves mutations of genes with roles in protein translation, histone methylation and blood coagulation. In terms of clinically relevant findings, unexpected activating mutations were found in the kinase BRAF, inhibitors of which have recently shown dramatic clinical activity. This suggests that BRAF inhibitors should be evaluated in patients with BRAF-mutated multiple myeloma. Multiple myeloma, a malignancy of plasma cells, remains incurable and is poorly understood. Using next-generation sequencing of several multiple myeloma genomes reveals that this disease involves mutations of genes involved in protein translation, histone methylation and blood coagulation. The study suggests that BRAF inhibitors should be evaluated in multiple myeloma clinical trials.

Recently, the advancement of deep learning (DL) in discriminative feature learning from 3-D LiDAR data has led to rapid development in the field of autonomous driving. However, automated processing uneven, unstructured, noisy, and massive 3-D point clouds are a challenging and tedious task. In this article, we provide a systematic review of existing compelling DL architectures applied in LiDAR point clouds, detailing for specific tasks in autonomous driving, such as segmentation, detection, and classification. Although several published research articles focus on specific topics in computer vision for autonomous vehicles, to date, no general survey on DL applied in LiDAR point clouds for autonomous vehicles exists. Thus, the goal of this article is to narrow the gap in this topic. More than 140 key contributions in the recent five years are summarized in this survey, including the milestone 3-D deep architectures, the remarkable DL applications in 3-D semantic segmentation, object detection, and classification; specific data sets, evaluation metrics, and the state-of-the-art performance. Finally, we conclude the remaining challenges and future researches.

The completion of whole genome sequencing projects has provided the genetic instructions of life. However, whereas the identification of gene coding regions has progressed, the mapping of transcriptional regulatory motifs has moved more slowly. To understand how distinct expression profiles can be established and maintained, a greater understanding of these sequences and their trans-acting factors is required. Herein we have used a combined in silico and biochemical approach to identify binding sites [repressor element 1/neuron-restrictive silencer element (RE1/NRSE)] and potential target genes of RE1 silencing transcription factor/neuron-restrictive silencing factor (REST/NRSF) within the human, mouse, and Fugu rubripes genomes. We have used this genome-wide analysis to identify 1,892 human, 1,894 mouse, and 554 Fugu RE1/NRSEs and present their location and gene linkages in a searchable database. Furthermore, we identified an in vivo hierarchy in which distinct subsets of RE1/NRSEs interact with endogenous levels of REST/NRSF, whereas others function as bona fide transcriptional control elements only in the presence of elevated levels of REST/NRSF. These data show that individual RE1/NRSE sites interact differentially with REST/NRSF within a particular cell type. This combined bioinformatic and biochemical approach serves to illustrate the selective manner in which a transcription factor interacts with its potential binding sites and regulates target genes. In addition, this approach provides a unique whole-genome map for a given transcription factor-binding site implicated in establishing specific patterns of neuronal gene expression.