Abstract. Deep convolutional networks have achieved great success for visual recognition in still images. However, for action recognition in videos, the advantage over traditional methods is not so evident. This paper aims to discover the principles to design effective ConvNet architectures for action recognition in videos and learn these models given limited training samples. Our first contribution is temporal segment network (TSN), a novel framework for video-based action recognition. which is based on the idea of long-range temporal structure modeling. It combines a sparse temporal sampling strategy and video-level supervision to enable efficient and effective learning using the whole action video. The other contribution is our study on a series of good practices in learning ConvNets on video data with the help of temporal segment network. Our approach obtains the state-the-of-art performance on the datasets of HMDB51 (69.4%) and UCF101 (94.2%). We also visualize the learned ConvNet models, which qualitatively demonstrates the effectiveness of temporal segment network and the proposed good practices.

The antimicrobial peptide database (APD, http://aps.unmc.edu/AP/) is an original database initially online in 2003. The APD2 (2009 version) has been regularly updated and further expanded into the APD3. This database currently focuses on natural antimicrobial peptides (AMPs) with defined sequence and activity. It includes a total of 2619 AMPs with 261 bacteriocins from bacteria, 4 AMPs from archaea, 7 from protists, 13 from fungi, 321 from plants and 1972 animal host defense peptides. The APD3 contains 2169 antibacterial, 172 antiviral, 105 anti-HIV, 959 antifungal, 80 antiparasitic and 185 anticancer peptides. Newly annotated are AMPs with antibiofilm, antimalarial, anti-protist, insecticidal, spermicidal, chemotactic, wound healing, antioxidant and protease inhibiting properties. We also describe other searchable annotations, including target pathogens, molecule-binding partners, post-translational modifications and animal models. Amino acid profiles or signatures of natural AMPs are important for peptide classification, prediction and design. Finally, we summarize various database applications in research and education.

The JUNO experiment locates in Jinji town, Kaiping city, Jiangmen city, Guangdong province. The geographic location is east longitude 112 • 31'05' and North latitude 22 • 07'05'. The experimental site is 43 km to the southwest of the Kaiping city, a county-level city in the prefecture-level city Jiangmen in Guangdong province. There are five big cities, Guangzhou, Hong Kong, Macau, Shenzhen, and Zhuhai, all in ∼200 km drive distance, as shown in figure 3.

3D object detection is receiving increasing attention from both industry and academia thanks to its wide applications in various fields. In this paper, we propose the Point-Voxel Region based Convolution Neural Networks (PV-RCNNs) for accurate 3D detection from point clouds. First, we propose a novel 3D object detector, PV-RCNN-v1, which employs the voxel-to-keypoint scene encoding and keypoint-to-grid RoI feature abstraction two novel steps. These two steps deeply incorporate both 3D voxel CNN and PointNet-based set abstraction for learning discriminative point-cloud features. Second, we propose a more advanced framework, PV-RCNN-v2, for more efficient and accurate 3D detection. It consists of two major improvements, where the first one is the sectorized proposal-centric strategy for efficiently producing more representative and uniformly distributed keypoints, and the second one is the VectorPool aggregation to replace set abstraction for better aggregating local point-cloud features with much less resource consumption. With these two major modifications, our PV-RCNN-v2 runs more than twice as fast as the v1 version while still achieving better performance on the large-scale Waymo Open Dataset with 150m × 150m detection range. Extensive experiments demonstrate that our proposed PV-RCNNs significantly outperform previous state-of-the-art 3D detection methods on both the Waymo Open Dataset and the highly-competitive KITTI benchmark.

Interest in two-dimensional (2D) van der Waals materials has grown rapidly across multiple scientific and engineering disciplines in recent years. However, ferroelectricity, the presence of a spontaneous electric polarization, which is important in many practical applications, has rarely been reported in such materials so far. Here we employ first-principles calculations to discover a branch of the 2D materials family, based on In2Se3 and other III2-VI3 van der Waals materials, that exhibits room-temperature ferroelectricity with reversible spontaneous electric polarization in both out-of-plane and in-plane orientations. The device potential of these 2D ferroelectric materials is further demonstrated using the examples of van der Waals heterostructures of In2Se3/graphene, exhibiting a tunable Schottky barrier, and In2Se3/WSe2, showing a significant band gap reduction in the combined system. These findings promise to substantially broaden the tunability of van der Waals heterostructures for a wide range of applications.