Nanjing University of Information Science and Technology

About: Nanjing University of Information Science and Technology is a education organization based out in Nanjing, China. It is known for research contribution in the topics: Precipitation & Aerosol. The organization has 14129 authors who have published 17985 publications receiving 267578 citations. The organization is also known as: Nan Xin Da.

Learning deep and wide: a spectral method for learning deep networks.

Abstract: Building intelligent systems that are capable of extracting high-level representations from high-dimensional sensory data lies at the core of solving many computer vision-related tasks. We propose the multispectral neural networks (MSNN) to learn features from multicolumn deep neural networks and embed the penultimate hierarchical discriminative manifolds into a compact representation. The low-dimensional embedding explores the complementary property of different views wherein the distribution of each view is sufficiently smooth and hence achieves robustness, given few labeled training data. Our experiments show that spectrally embedding several deep neural networks can explore the optimum output from the multicolumn networks and consistently decrease the error rate compared with a single deep network.

Characteristic, changes and impacts of permafrost on Qinghai-Tibet Plateau

Abstract: Qinghai-Tibet Plateau (QTP) is the largest high-altitude permafrost zone in the mid-latitudes. Due to the climate warming, permafrost degradation on the QTP has been widely recorded in the past decades. Since it greatly affects the East Asian monsoon, and even the global climate system, it is extremely important to understand permafrost current state, changes and its impacts. Based on literature reviews and some new data, this paper summarizes the characteristics of the current state permafrost on the QTP, including the active layer thickness (ALT), the spatial distribution of permafrost, permafrost temperature and thickness, as well as the ground ice and soil carbon storage in permafrost region. The new results showed that the permafrost and seasonally frozen ground area (excluding glaciers and lakes) is 1.06 million square kilometers and 1.45 million square kilometers on the QTP. The sub-stable, transitional, and unstable permafrost accounts for 30.4%, 22.1% and 22.6% of the total permafrost area. The permafrost thickness varies greatly among topography, with the maximum value in mountainous areas, which could be deeper than 200 m, while the minimum value in the flat areas and mountain valleys, which could be less than 60 m. The mean active layer thickness of the permafrost on the QTP is 2.3 m, with 80% of the permafrost regions ranges from 0.8 m to 3.5 m. During 1980 to 2015, soil temperature at 0−10, 10−40, 40−100, 100−200 cm increased at a rate of 0.439, 0.449, 0.396, and 0.259°C/ 10 a, respectively. From 2004 to 2018, the increasing rate of the soil temperature at the bottom of active layer was 0.486°C/ 10 a. These results show that the permafrost degradation has been accelerating. The permafrost degradation largely reduces the soil moisture. The ground ice volume of the permafrost is estimated up to 1.27×104 km3 (liquid water equivalent). The soil organic carbon in the upper 2 m of permafrost region is about 17 Pg; there is a large uncertainty in this estimation however due to the great heterogeneities in the soil column. Although the permafrost ecosystem is a carbon sink at the present, it is possible that it will shift to a carbon source due to the loss of soil organic carbon along with permafrost degradation. Overall, this paper shows that the plateau permafrost has undergone remarkable degradation during past decades, which are clearly proven by the increasing ALTs and ground temperature. Most of the permafrost on the QTP belongs to the unstable permafrost, meaning that permafrost over TPQ is very sensitive to climate warming. The permafrost interacts closely with water, soil, greenhouse gases emission and biosphere. Therefore, the permafrost degradation greatly affects the regional hydrology, ecology and even the global climate system. This paper also proposes approaches and methods to study the interactive mechanisms between permafrost and climate change, and the results can serve as a scientific basis for environmental protection, engineering design and construction in cold regions.

Energy-efficient cluster-based dynamic routes adjustment approach for wireless sensor networks with mobile sinks

Abstract: In wireless sensor networks (WSNs), sensor nodes near static sink will have more traffic load to forward and the network lifetime will get largely reduced. This problem is referred to as the hotspot problem. Recently, adopting sink mobility has been considered as a good strategy to overcome the hotspot problem. Despite its many advantages, due to the dynamic network topology caused by sink mobility, data transmission to the mobile sink is a challenging task. To achieve efficient data dissemination, nodes need to reconstruct their routes toward the latest location of the mobile sink, which weakens the energy conservation aim. In this paper, we proposed an energy-efficient cluster-based dynamic routes adjustment approach (EECDRA) which aims to minimize the routes reconstruction cost of the sensor nodes while maintaining nearly optimal routes to the latest location of the mobile sinks. The network is divided into several equal clusters and cluster heads are selected within each cluster. We also set some communication rules that manage routes reconstruction process accordingly requiring only a limited number of nodes to readjust their data delivery routes toward the mobile sinks. Simulation results show that the mobile sinks for reducing reconstruction of route have improved the energy efficiency and prolonged lifetime of wireless sensor network.

Anomaly Detection Based on Convolutional Recurrent Autoencoder for IoT Time Series

Abstract: Internet of Things (IoT) realizes the interconnection of heterogeneous devices by the technology of wireless and mobile communication. The data of target regions are collected by widely distributed sensing devices and transmitted to the processing center for aggregation and analysis as the basis of IoT. The quality of IoT services usually depends on the accuracy and integrity of data. However, due to the adverse environment or device defects, the collected data will be anomalous. Therefore, the effective method of anomaly detection is the crucial issue for guaranteeing service quality. Deep learning is one of the most concerned technology in recent years which realizes automatic feature extraction from raw data. In this article, the integrated model of the convolutional neural network (CNN) and recurrent autoencoder is proposed for anomaly detection. Simple combination of CNN and autoencoder cannot improve classification performance, especially, for time series. Therefore, we utilize the two-stage sliding window in data preprocessing to learn better representations. Based on the characteristics of the Yahoo Webscope S5 dataset, raw time series with anomalous points are extended to fixed-length sequences with normal or anomaly label via the first-stage sliding window. Then, each sequence is transformed into continuous time-dependent subsequences by another smaller sliding window. The preprocessing of the two-stage sliding window can be considered as low-level temporal feature extraction, and we empirically prove that the preprocessing of the two-stage sliding window will be useful for high-level feature extraction in the integrated model. After data preprocessing, spatial and temporal features are extracted in CNN and recurrent autoencoder for the classification in fully connected networks. Empiric results show that the proposed model has better performances on multiple classification metrics and achieves preferable effect on anomaly detection.

A survey of decision-making methods with probabilistic linguistic information: bibliometrics, preliminaries, methodologies, applications and future directions

Abstract: Probabilistic linguistic term set (PLTS) has been proposed to tackle qualitative information efficiently in the decision-making process to achieve computing with expressions, which can be regarded as an advanced process of computing with words. The PLTS plays an important role in decision making by providing a comprehensive way for representing complex linguistic information. Owning to its usefulness and efficiency, the PLTS has attracted a lot of researchers’ attention, and fruitful research achievements regarding it has been published since it was introduced in 2016. As the probabilistic linguistic theory is still in its infancy, a survey of it contributes to understanding the previous topics and the current issues, and predicting the future research directions in this area. To implement these goals, this paper is organized by bibliometric analysis, definitions, multiple criteria decision-making methods, preference relations, and applications. Twelve future research directions related to the probabilistic linguistic decision-making theory are indicated. This paper provides some insights for researchers and practitioners who have interest in complex linguistic decision making.