Search

We develop online scheduling policies to minimize the sum average age of information (AoI) subject to transmission capacity and long-run average resource constraints in a multisource two-hop system, where independent sources randomly generate status update packets which are sent to the destination via a relay through error-prone links. A stochastic optimization problem is formulated and solved in known and unknown environments. For the known environment, an online nearoptimal low-complexity policy is developed using the driftplus-penalty method. For the unknown environment, a deep reinforcement learning policy is developed by employing the Lyapunov optimization theory and a dueling double deep Qnetwork. Simulation results show up to 136% performance improvement of the proposed policy compared to a greedy-based baseline policy.

At present, most of the methods used in the research of image semantic segmentation ignore the low-level feature information of image, such as space, edge, etc., which leads to the problems that the segmentation of edge and small part is not precise enough and the accuracy of segmentation result is not high. To solve this problem, this paper proposes a multi-source fusion image semantic segmentation model of generative adversarial networks based on FCN: SCAGAN. In VGG19 network, add super-pixel and edge detection algorithm, and introduce the efficient spatial pyramid module to reduce the number of parameters while adding the spatial and edge information of image; Adjust the skipping structure to better integrate the low-level features and high-level features; build a generation model DeepLab-SCFCN combining with the atrous spatial pyramid pooling to better capture the feature information of different scales of the target for segmentation; The FCN with five modules is designed as the discrimination model for GAN. It is verified on the data set PASCAL VOC 2012 that the model achieves IoU of 70.1% with a small number of network layers, and the segmentation effect of edge and small part is better at the same time. This technology can be used in image semantic segmentation.

Current iris recognition technology faces practical difficulties. For example, due to the unsteady morphology of a heterogeneous iris generated by a variety of different devices and environments, the traditional processing methods of statistical learning or cognitive learning for a single iris source are not effective. The existing iris data set size and situational classification constraints make it difficult to meet the requirements of learning methods under a single deep learning framework. Therefore, this paper proposes a method of heterogeneous iris recognition based on an entropy feature lightweight neural network under multi-source feature fusion. The method is divided into an image-processing module and a recognition module. The image-processing module converts the iris image into a recognition label via a convolutional neural network. The recognition module is based on statistical learning ideas and design of a multi-source feature fusion mechanism. The information entropy of the iris feature label is used to set the iris entropy feature category label and design the recognition function according to the category label to obtain the recognition result. As the requirement for the number and quality of irises changes, the category labels in the recognition function are dynamically adjusted using a feedback learning mechanism. This paper uses iris data collected from three different devices in the JLU iris library. The experimental results prove that for multi-category classification of lightweight constrained multi-state irises, the abovementioned problems are ameliorated to a certain extent by this method.

Accurate estimation of aboveground forest biomass (AGB) at a large scale is important in global carbon cycle, forest productivity, and climate change. Coarse resolution remote sensing data of long time series are often used to estimate large scale AGB, but the result is inaccurate due to the scaling effect caused by nonlinearity in data representation and the existence of mixed pixels containing different forest types and land uses. Improvement in the accuracy of AGB estimated from coarse resolution remote sensing data is urgently needed. Research on spatial scaling of AGB is still lacking, therefore, this article proposed an approach based on structural analysis of mixed pixels and the Random Forest model (SMPRF) to increase the accuracy of AGB estimated from coarse resolution data. MODIS and SPOT 5 data were used to create forest biomass distribution maps of the study area at two scales. The scaling effect on estimating forest biomass based on remote sensing was analyzed by comparing data from these two datasets. SMPRF, which included a correction factor for the scaling effect on AGB estimated from coarse resolution MODIS data, was used to create a model that scaled from the fine resolution data (SPOT 5) to the coarse resolution data (MODIS). The results showed that the accuracy of AGB estimated from MODIS data was increased using this method. The Pearson correlation coefficient ( $r$ ) for data verification increased from 0.63 to 0.89 and the root mean squared error decreased from 51.6 Mg $\cdot $ ha−1 to 26.8 Mg $\cdot $ ha−1. The difference tests showed that the changes were extremely significant ( $p = 0$ ). Thus, SMPRF can significantly improve the accuracy of large scale AGB estimation based on coarse resolution remote sensing data and the feasibility of applying the method proposed in this study to related fields is verified.

The harmonic reducer of the space drive mechanism is one of the critical components of the satellite remote sensing system and its reliability is always required strictly in practice. During the manufacturing and usage processes of the harmonic reducer, uncertainties are ubiquitous and performance degradation is inevitable, which is mainly caused by the wear between the meshed tooth surfaces. They influence the reliability of harmonic reducer. In this paper, a performance margin modeling and reliability analysis for harmonic reducer considering multi-source uncertainties and wear is proposed. Firstly, based on the functional principle and the influence of wear, the performance margin model is established using hysteresis and transmission errors as key performance parameters. Then multi-source uncertainties are analyzed and quantified, including manufacturing errors, uncertainties in operational and environmental conditions, and uncertainties in performance thresholds. Finally, the reliability model is constructed based on the performance margin. A case study of a harmonic reducer is applied and the reliability sensitivity analysis is implemented to show the practicability of the proposed method. The results show that the proposed method can provide some suggestions to the design and manufacturing phases of the harmonic reducer.

Automatic vehicle identification (AVI) data, Integrated Circuit (IC) card data and Global Positioning System (GPS) data offer an emerging and promising source of information for analysis of traffic problems. Research on insights and information from AVI data for transport analysis has made little progress in developing specific applications especially. The emergence of multi-source data provides us with a new perspective for multi-mode transportation. This paper proposes a multi-mode traffic demand forecasting method based on AVI data, metro IC card data, and taxi GPS data. The paper extracts traffic origins and destinations (OD) information of travelers from the multi-source data and uses the extracted data for traffic zone division. Finally, a multi-mode traffic forecasting model is established on this basis. GPS data of taxi trips are selected as the clustering data and k-means algorithm was adopted to divide traffic zones in Shenzhen. Moreover, the research applies the principle of convex hull to outline the boundary of the cell. Additionally, this paper establishes the multi-mode transportation forecasting model by integrating the correlation between various transportation modes into the deep learning model for prediction. The results show that the multi-mode demand forecasting model has higher accuracy and better forecasting results comparing it with the single-mode demand forecasting model which is referring to the conventional four-step procedure. The result demonstrates that effective traffic and travel data can be obtained from multi-source data, providing an opportunity to improve the analysis of complex travel patterns and behaviors for travel demand modeling and transportation planning. Furthermore, the substantive contribution of this research is that it provides strong empirical evidence for the existence of correlation among multi-mode travels and travel demand.

Mining a document structure from multiple data sources in terms of their underlying topics has become an important task of document clustering. The traditional document clustering approach cannot be applied directly to the multi-source document clustering problem. There are three typical difficulties: 1) The topics of different data sources are related but not the same. 2) Usually, each data source has its own focus on topics. 3) The number of clusters of the data sources are not necessarily the same and are not known beforehand. In this paper, based on our previous research, we design a novel multi-source document clustering model, namely, the hierarchical Dirichlet multinomial allocation (HDMA) model, to solve all the above problems. The HDMA model is investigated with a two-step hierarchical topic generation process. Topics are learnt to share their general characteristics across data source, while at the same time preserve the local characteristics of the data source. Each data source is applied with an exclusive topic partition to learn the source-level topic emphasis. A Gibbs sampling algorithm is then used to learn the number of clusters for each data source as well as the parameters of the HDMA model at the same time. Experimental results demonstrate that the HDMA model is effective.

Live holographic teleportation is an emerging media application that allows Internet users to communicate with each other in a fully immersive manner. One distinct feature of such an application is the capability of simultaneously teleporting multiple objects from different network locations to the receiver’s field of view, mimicking the effect of group-based communications in a common physical space. In this case, teleportation frames from individual sources need to be stringently synchronized in order to assure user Quality of Experiences (QoE) in terms of avoiding the perception of motion misalignment at the receiver side. In this paper, we carry out systematic performance evaluations on how different Internet path conditions may affect the teleportation frame synchronisation performances. Based on this, we present a lightweight, edge-computing based scheme that is able to achieve controllable frame synchronisation operations for multi-source based teleportation applications at the Internet scale.

DC voltage has been targeted for electric transportation where failures can impact the safety of passengers or may jeopardize the success of a mission and thus are unacceptable. Because electrical insulation determines the life of an electrical system, monitoring the aging of insulation systems is essential. In this regard, partial discharge (PD) activity can be used as a primary indicator in monitoring the weakness of insulation materials. In the power corridor of electric transportation, there are many electrical generation, distribution, and storage apparatuses that each are a PD source and the input PD signal measured is a mixture of all PD sources. Separating and recognizing multiple PD sources has become a hot and, of course, challenging research topic in the dielectrics and electrical insulation area; deep learning (DL) is being considered as a promising technique to deal with it.

This paper presents the preliminary results on the use of Machine Learning (ML) for the estimation of the electric-field exposure in indoor scenarios with multiple WiFi sources. Differently from similar previous approaches, the present approach aims to design a Neural Network (NN) capable to address complex indoor scenarios that include not only down-link transmission by access points (APs) but also up-link transmission by several clients (e.g., laptop, printers, tablets, and smartphones). The NN was trained and tested on the field generated by multiple WiFi sources (2400 MHz) in an office indoor setup; the ‘target’ exposure field in such a scenario was derived using a deterministic indoor network planner method. The median prediction accuracy of the ‘target’ field exposure by the proposed NN was 0.0 dB (1st quartile: −0.7 dB; 3rd quartile 0.9 dB), with a root mean square error of 2.1 dB. The proposed approach is fast (the NN training lasts about 30 minutes) and could be useful to assess radio-frequency (RF) exposure in complex indoor scenarios.

Virtual panorama technology based on multi-source information fusion is a virtual reality technology based on image technology to generate realistic graphics. Panorama production software is used to generate panoramic images with a strong sense of scene and perspective effect that can be fully displayed at 720 degrees. Various kinds of voice, video, image and other hot spots are added in the virtual panorama image, and the step and front-end customized display are freely added, and multiple types of artificial intelligence graphic algorithms are applied to realize the virtual panorama platform of multi-source information fusion. The technology has been widely used in some businesses of the Yellow River, with strong practicability, large application and development space, and obvious advantages in investment benefit ratio.

The health condition of industrial equipment is closely related to productivity and safety, attaching great importance to fault diagnosis. Although current fault diagnosis methods have already achieved good effect to some degree, it is more reliable and outstanding to utilize multiple signal sources and pay attention to the effective information differences between various signals. In order to take full advantage of fault information, this paper proposes a new method of fault diagnosis based on multi-source signals and attention mechanism. This method learns the weights of multi-source features with attention mechanism and recalibrates the feature responses. Besides it extracts and fuses the fault features based on residual network (ResNet). The experimental results show that the proposed method can effectively learn the differences of multisource signals, and has excellent classification performance compared to those based on a single signal or multiple signals without attention mechanism.

Analytical modeling of Age of Information (AoI) and Peak AoI (PAoI) has recently drawn a lot of attention in the context of quantitative assessment of information freshness in status update systems. In this paper, we study a probabilistically preemptive bufferless M/PH/1/1 queue fed with information update packets from $N$ separate information sources for which a new information packet arrival from source- $m$ is allowed to preempt a packet from source- $n$ in service, with a probability depending on $n$ and $m$ . To make the model even more general than the existing ones, we assume a distinct phase-type (PH-type) service time distribution, a distinct packet error and retransmission probability, for each of the information sources. Using sample path arguments and the theory of Markov Fluid Queues (MFQ), the exact distributions of the AoI and PAoI are numerically obtained for each of the sources. Numerical examples are provided to demonstrate the impact of various system parameters on AoI performance. In the context of a two-source system, we present a methodology on how to optimally choose the preemption probabilities and packet generation rates so as to minimize certain AoI-oriented cost functions.

This paper presents a modular approach to model and control multi-source systems or networks. This method derives from an energy-based modeling approach called functional modeling in which predictive functional control algorithms have been implemented. A concept of cost function is introduced and associated to each source. The design of the control has several objectives: remain as generic as possible, satisfy the needs from the consumers and distribute flows in order to minimize the cost functions. This methodology is applied to a use case consisting in the management of an isolated production unit composed of a wind power plant and a backup battery. The system is modeled and controlled in order to ensure the active power balance between consumers needs and sources supplies.

Partial discharge (PD) is a major aging factor in insulation systems. The first step toward ensuring the health of insulation systems is the identification of PD sources. This would enable the system operator to take proper actions before the occurrence of the final breakdown. This fundamental study targets building and training a deep residual neural network (ResNet) model to detect and discriminate single and multiple sources of corona discharge that are acting simultaneously. The input of the ResNet model is the actual PD measurements under atmospheric pressure for seven electrode configurations: four single-source and three double-source configurations which expands from weakly nonuniform to extremely nonuniform electric field distribution. In this study, the measurement data are converted into standardized phase-resolved PD (PRPD) images that would be further used as input to the optimized ResNet model. The validation of the 62-layer deep neural network is evaluated through 30% randomly chosen images from the original dataset.

Aiming at the problem that traditional early warning systems ignore the multi-source nature of rainfall data, leading to poor geological hazard early warning, a geological hazard early warning system based on multi-source rainfall data is proposed and designed. This system includes multiple monitoring subsystems, mainly rainwater, early warning, etc., combined with early warning indicators to give the system's power supply module, temperature and humidity module, processor circuit, rain power supply, liquid level sensor circuit and other hardware structures; in software design In the process, based on the weighted moving average combined with the linear system, the filter mathematical model is established to analyze the relationship between the accumulated rainfall and the rainfall intensity, and the multi-source rainfall data statistics are realized. Experimental results show that the use of improved design early warning system can effectively improve the accuracy of early warning, enhance the effect of early warning, and the early warning results are more in line with actual conditions.

The traditional power grid disaster recovery plan has been unable to adapt to today's multi-source feeding new power grid structure. Based on the indicators of the traditional disaster recovery plan, this paper adds new indicators that adapt to the characteristics of the new power grid structure, and builds a new disaster recovery plan evaluation system that adapts to the contemporary power grid. It also proposes a comprehensive weight modeling method that combines the D-AHP and Poisson distribution methods to determine the index weight and time weight. Finally, the TOPSIS method is used to solve the problem of a new regional power grid. Sort the programs. The results of calculation examples show that the method proposed in this paper can effectively evaluate the scientific nature of the new power grid disaster recovery plan.

the spatial distribution and regulation characteristics of renewable energy are analyzed separately, and by analyzing the peaking demand of renewable energy access to Xiajiu area, a multi-source coordinated peak-shaving strategy is proposed. Firstly, the equivalent load curve and hydropower output curve are used to determine the starting point of hydropower peaking. The peaking demand, the optimization model with the largest peaking demand is established, and the system optimization model is calculated by using the branch and bound algorithm to determine the unit combination scheme of the system.

The ongoing COVID-19 pandemic has overloaded current healthcare systems, including radiology systems and departments. Machine learning-based medical imaging diagnostic approaches play an important role in tracking the spread of this virus, identifying high-risk patients, and controlling infections in real-time. Researchers aggregate radiographic samples from different data sources to establish a multi-source learning scheme to mitigate the insufficiency of COVID-19 samples from individual hospitals, especially in the early stage of the disease. However, data heterogeneity across different clinical centers with various imaging conditions is considered a significant limitation in model performance. This paper proposes a contrastive learning scheme for the automatic diagnosis of COVID-19 to effectively mitigate data heterogeneity in multi-source data and learn a robust and generalizable model. Inspired by advances in domain adaptation, we employ contrastive training objectives to promote intra-class cohesion across different data sources and inter-class separation of infected and non-infected cases. Extensive experiments on two public COVID-19 CT datasets demonstrate the effectiveness of the proposed method for tackling data heterogeneity problems with boosted diagnosis performance. Moreover, benefiting from the contrastive learning framework, our method can be generalized to solve data heterogeneity problems under a broader multi-source learning setting.

Fetching different parts of the same content (file) simultaneously through multiple network paths has been found to improve content delivery. There are several application layer programs that use this technique to improve the perceived performance at the end users. However, these applications use multiple sockets (multiple connections) at the transport layer, which has several disadvantages. Also, the existing transport layer protocols that allow content delivery over multiple network paths over a single transport layer connection (e.g., MPTCP) are limited to content delivery from a single source. With the availability of content across distributed content servers, there is a need for a transport layer protocol that provides the ability to deliver content from these distributed sources over a single transport layer connection. In this paper, we design and implement a multi-source transport control protocol (MSTCP) that can be used to deliver content from a distributed source to a client application over a single transport layer connection. A prototype implementation and preliminary performance measures showing the effectiveness of MSTCP are also presented.

The quality of sinter ore is a key part in ironmaking industry, and the content of FeO of sinter ore is one of its significant quality indexes. In order to predict the content of FeO of ore agglomerate, a data-driven prediction scheme based on multi-source information and long short-term memory (Long Short-Term Memory, LSTM) network is proposed. In the proposed method, the multi-source information is firstly extracted and then they are input to the LSTM, whose target output is the corresponding Reference-FeO. The multi-source information used in this paper includes image, vibration and temperature information. And for the feature extraction of the images, a method is developed to preprocess and extract features from images. Besides, the information of vibration and temperature are selected on account of the correlation analysis and mechanism of sintering. Moreover, the corresponding Reference-FeO at the end of the sintering pallets is obtained as the target output of the LSTM network. Finally, a LSTM network is utilized to predict the target output. The experimental results show that the data-driven prediction scheme using multi-source features can reach a good performance with absolute error within 0.05 and can meet the needs of practical engineering.

The safety of the equipment in the communication room is the basis to ensure the reliable operation of the urban rail transit. Therefore, it is of great significance to investigate the complex electromagnetic field distribution and radiation level in the communication room. On the basis of investigation, the typical interference sources and their transmission routes are explored first. The influence of each interference source on the internal electromagnetic environment is simulated and analyzed, which shows good consistency with the measured.

With the rapid development of high-speed trains in the world, how to quickly collect train operation data becomes a significant issue. The paper uses Global Navigation Satellite System (GNSS), Inertial Navigation System (INS), and Driver Machine Interface (DMI) information fusion to obtain the position and speed information of the train more accurately. Among these information sources, GNSS and INS can carry out the collection of train position and speed information with and without obstructions, respectively. When the train is running in the open and unobstructed, GNSS can directly obtain train operation information. When the train passes through an environment with obstructions, GNSS and INS work simultaneously to obtain train operation information. The DMI on the train has the timely speed and kilometer mark information of the train. During the measurement, as we can not access to the onboard equipment. The camera is applied to collect the DMI interface. The image is processed to obtain the train data information. In this paper, the train data collected by GNSS/INS equipment is corrected by the Kalman filtering method. The data collected by DMI is filtered by the kinematics theorem and the least square method. What is more, a centralized fusion structure is used to process the DMI, GNSS, and INS data. This result can provide more accurate train time, speed, and kilometer mark information for the train during the operation test.

Muscle fatigue detection can be of good help to many tasks such as athletes’ physical training and soldiers’ body status monitoring. Surface elecrtromyography (sEMG) signals are widely used in muscle fatigue detection. However, sEMG signals exist only when the muscle contracts and disappear when it relaxes, making muscle fatigue detection methods cannot work well in realistic applications. To solve this problem, a method based on phase space reconstruction is proposed to automatically filter useless signals and retain useful ones from raw sensor data, improving the practicality of the detection methods. In previous works on muscle fatigue detection, most researchers took only sEMG signals of the target muscle into consideration. However, in reality, when someone is doing physical work, several cooperative muscles rather than some single one participate in the task. Therefore, the exercise status of one muscle not only resides in its own sEMG signals, but also is included in its partners’. For this reason, a fatigue detection method to muscle fatigue detection based on integrating multi-source sEMG signals is proposed, where long short-term memories (LSTM) and one attention layer are used as an inference model. Moreover, a series of sequential detection results are integrated to make a final result to deal with accidental wrong judgements, which further improves the practicality. In our experiments, our LSTM-Attention-based method achieves an detection accuracy of 90.4%, which is much better than the method based on LSTM processing sEMG signals only from the target muscle.

With the development of information technology, in order to improve the accuracy of target recognition, information collaboration technology plays an increasingly important role. However, cooperating with ISAR images and infrared images for target recognition is invisible. In this paper, we propose for the first time to fuse the sample feature information of ISAR images and infrared images for target recognition. Initially, the same preprocessing and feature extraction are performed on ISAR and infrared images respectively. Subsequently, their own features are weighted by the weight coefficient to complete the feature-level fusion. Finally, BP neural network is used for image classification. The experimental results confirm that the multi-source collaborative classification has higher accuracy than single-source.

In current scenario, transfer learning or domain adaptation has been emerged as fruitful approach to handle problems of distribution mismatch between the training and test data. Standard machine learning techniques utilize labeled data for getting better performance. In practical scenario, due to scarcity of labeled data, the classifier trained on the source domain cannot be utilized efficiently for classifying the target domain data. So, there is a need to use the previously acquired knowledge from a related source domain to classify the information in the target domain. Previous works on single/multiple source domain adaptation have achieved substantial growth in tackling this concern. In this work, we have proposed a novel unsupervised multi-source based approach MSVDA for visual domain adaptation in which data from multiple labeled source domains are utilized to classify the information in the target domain containing unlabeled data. Our proposed approach MSVDA extends the existing Maximum Mean Discrepancy criteria across various source domains and also preserves the discriminative information of various source domains. Further, it learns an optimal classification that diminishes the empirical risk and enhances the consistency rate between the prediction function and the manifold. Various experiments on the two publicly available datasets with multiple-domain scenario settings (double-domain, triple-domain and quadruple-domain scenarios) have demonstrated the efficacy of our proposed method MSVDA over other existing methods.

With the diversified sources satisfying users' needs, many online service platforms collect information from multiple sources in order to provide a set of useful information to the users. However, existing recommendation systems are mostly designed for single-source data, and thus fail to recommend multi-source review packages since the interplay between the reviews of different sources is not properly modeled. In fact, modeling the interplay between different sources is challenging because 1) two reviews may conflict with each other, 2) different users have different preferences on review sources, and 3) users' preferences to each source may change under different scenarios. To address these challenges, we propose Self-Attentive Recommendation for multi-source review Package (SARP), for predicting how useful the user feels to the package, while simultaneously reflecting how much the user is affected by each review. Specifically, SARP jointly considers the relationships of every user, purpose, and review source to learn better latent representations. A self-attention module is further used for integrating source representations and the review ratings, following a multi-layer perceptron (MLP) for the prediction tasks. Experimental results on the self-constructed dataset and public dataset demonstrate that the proposed model outperforms the state-of-the-art approaches.

Mesoscale eddies are swirling water exiting ubiquitously in the global ocean. They are significant for material and energy transit and global ocean circulation. Using satellite sea surface height anomaly (SSHA) and sea surface temperature (SST) images, this study proposes a new automated eddy tracking model based on deep learning (DL) technology. Compared with existing eddy tracking algorithms, the DL-based model fuses SSHA and SST data for eddy tracking. We aim to solve problems as eddy splits, mergers, and transient ‘disappearance’ that occurred using SSHA data. The DL-based eddy tracking model is applied in the Kuroshio Extension, and the result verified the model's tracking accuracy and efficiency.

Recently, the image segmentation has been a significant research direction in the field of optical remote sensing data processing. However, due to the limitation of the optical imaging mechanism, traditional image segmentation methods are not efficient for processing the optical remote sensing images, especially influencing by the complex weather conditions. In order to ensure the classification performance, synthetic aperture radar (SAR) data are employed as complementary to the data procedure for enhancing the capability of land cover interpretation. Then a dual-stream high-resolution network (HRNet) is proposed to combine two types of heterogeneous data (SAR and optical image), and a multi-modal squeeze-and-excitation (SE) module is exploited to make feature maps fused. Experiments show that the proposed method has excellent performance on the remote sensing data acquired by GF2 and GF3 satellites.

Forest Dead Fuel Load (FDFL) is vital for fire prevention and suppression since it affects the surface fire ignition and intensity directly. The accurate spatial distribution information of FDFL can provide decision support for fire managers. Remote sensing (RS) technology is a unique way to estimate FDFL on a large scale. However, most researches focus on the application of LiDAR data which is expensive to analyze fuel dynamics on spatiotemporal scale. Little attention has been devoted to freely accessible datasets such as Sentinel-1 and Sentinel-2 with global coverage. This study not only combined these two data but also the site conditions (i.e., elevation, slope and aspect) to estimate the FDFL in the southwest of Sichuan, China. The machine learning method, Random Forest Regression (RFR) was selected to manage the multiple and nonlinear relationships between RS data and FDFL. Results show that 1h and all FDFL (the sum of 1h, 10h, 100h and litter) can be more indicated by RS data (1h: $\mathrm{R}^{2}=0.57,\ \text{RMSE}=0.18$ Tons/ha; All: $\mathrm{R}^{2}=0.59,\ \text{RMSE}=1.81$ Tons/ha). The representation ability of RS data for 10h, 100h and litter is relatively weaker (10h: $\mathrm{R}^{2}=0.41,\ \text{RMSE}=0.47$ Tons/ha; 100h: $\mathrm{R}^{2}=0.40,\ \ \text{RMSE}=1.07$ Tons/ha; litter: $\mathrm{R}^{2}=0.29,\quad \text{RMSE}=1.66$ Tons/ha). Hence, this study demonstrated the potential of multi-source RS data for FDFL estimation.

Satellite remote sensing is playing an increasing role in volcano monitoring, not only at little-known and remote volcanoes where in-situ data are often unavailable, but also at well-monitored volcanoes to complement ground-based measurements. The synoptic view captured by multi-source satellite imagery over volcanoes can benefit hazard monitoring efforts, both following the different phases and intensities of an eruption, as well as helping in nowcasting and eventually forecasting the areas potentially threatened by hazardous phenomena. Here we demonstrate how satellite data from different sources can be combined to refine the satellite-derived radiant heat flux and time-averaged discharge rates (TADRs) estimation, and to improve real-time monitoring and nowcasting of eruptive events. This is achieved by cross-referencing TADRs obtained from low spatial/high temporal resolution satellite data (e.g. MODIS, SEVIRI, VIIRS, SLSTR) with maps of volcanic deposits including the time-varying evolution of lava flow emplacement derived from medium spatial resolution multispectral satellite data (e.g. EO-ALI, Landsat, Sentinel-2, ASTER), and with detailed three-dimensional maps of volcanic deposits retrieved from the topographic monitoring by using stereo or tri-stereo optical data (e.g. Pleiades, PlanetScope, SkySat, ASTER). Finally, satellite-derived parameters are used as input and validation for the numerical modelling of lava flow scenarios. This strategy is here applied to recent eruptions occurred on Etna, Stromboli and Nabro volcanoes.

The wide deployment of surveillance systems has shown the necessity to recognize human activities in videos. Existing work achieved high recognition accuracy for single- person activities and some group activities. In this paper, we identify a more challenging issue: recognizing multiple co- evolving but asynchronous activities from a set of videos captured by multiple cameras that share overlapped views. To address this issue, we design a system M&M to fuse objects and 2D skeletons from multi-source videos to reconstruct the complete 3D view-invariant model of multiple activity scenarios, which enables accurate recognition of cross-source activities. By embedding the 3D model with a concise graph representation, we propose an efficient recognition method in a bottom-up manner to achieve high accuracy and good scalability to changing complexity of the captured scenario. We collect and release a dataset containing correlated multi-source videos for multiple co-evolving activities, and evaluate our design on it. Experimental results show that M&M achieves 91.2% average accuracy for 10 types of coevolving activities.

Improved Grey Wolf Optimizer (I-GWO) is a new explored to develop an algorithm for solving engineering issues. Recently, the I-GWO is very popular among researchers due to its effectiveness in solving complex engineering problems especially that related to power systems. This paper proposes I-GWO based automatic generation control for elimination of the frequency and power interchange deviation in an interconnected power system. A two area six-unit multi-source system equipped with PID and PID with derivative filter (PIDF) controllers is taken in consideration. I-GWO is used to seek for optimal parameters of controllers. The performance of the suggested optimal controller is compared and evaluated with that of the conventional PID controller in order to demonstrate the superior efficiency of the suggested I-GWO in PIDF controller tuning. The simulation results prove that the optimal I-GWO based controller provides better performance as compared with the conventional controller.