Search

In RNA-seq data processing, short reads are usually aligned from one species against its own genome sequence; however, in plant-pathogen interaction systems, reads from both host and pathogen samples are blended together. In contrast with single-genome analyses, both pathogen and host reference genomes are involved in the alignment process. In such circumstances, the order in which the alignment is carried out, whether the host or pathogen is aligned first, or if both genomes are aligned simultaneously, influences the read counts of certain genes. This is a problem, especially at advanced infection stages. It is crucial to have an appropriate strategy for aligning the reads to their respective genomes, yet the existing strategies of either sequential or parallel alignment become problematic when mapping mixed reads to their corresponding reference genomes. The challenge lies in the determination of which reads belong to which species, especially when homology exists between the host and pathogen genomes. This chapter proposes a combo-genome alignment strategy, which was compared with existing alignment scenarios. Simulation results demonstrated that the degree of discrepancy in the results is correlated with phylogenetic distance of the two species in the mixture which was attributable to the extent of homology between the two genomes involved. This correlation was also found in the analysis using two real RNA-seq datasets of Fusariumchallenged wheat plants. Comparisons of the three RNA-seq processing strategies on three simulation datasets and two real Fusarium-infected wheat datasets showed that an alignment to a combo-genome, consisting of both host and pathogen genomes, improves mapping quality as compared to sequential alignment procedures., Series: Methods in Molecular Biology (MIMB); no. 2659

Given a natural language description, text-based person retrieval aims to identify images of a target person from a large-scale person image database. Existing methods generally face a \textbf{color over-reliance problem}, which means that the models rely heavily on color information when matching cross-modal data. Indeed, color information is an important decision-making accordance for retrieval, but the over-reliance on color would distract the model from other key clues (e.g. texture information, structural information, etc.), and thereby lead to a sub-optimal retrieval performance. To solve this problem, in this paper, we propose to \textbf{C}apture \textbf{A}ll-round \textbf{I}nformation \textbf{B}eyond \textbf{C}olor (\textbf{CAIBC}) via a jointly optimized multi-branch architecture for text-based person retrieval. CAIBC contains three branches including an RGB branch, a grayscale (GRS) branch and a color (CLR) branch. Besides, with the aim of making full use of all-round information in a balanced and effective way, a mutual learning mechanism is employed to enable the three branches which attend to varied aspects of information to communicate with and learn from each other. Extensive experimental analysis is carried out to evaluate our proposed CAIBC method on the CUHK-PEDES and RSTPReid datasets in both \textbf{supervised} and \textbf{weakly supervised} text-based person retrieval settings, which demonstrates that CAIBC significantly outperforms existing methods and achieves the state-of-the-art performance on all the three tasks., Accepted on ACM MM '22

Recently Digital Twin (DT) has attracted much attention from researchers due to its capacity of system monitoring and health management to improve the reliability and availability of systems. This emerging technology has been considered a promising solution for various sectors to enhance the sustainability of business development. In general, DT relies on the living models for simulating system behaviors to monitor the systems or assets in production. Such models could be either mathematical/physics-based or data-driven, which are able to explain, predict, and describe system behaviors timely and accurately. Therefore, the machine learning-enabled modeling technology has become a powerful tool to develop such data-driven living models. However, data-driven models developed with supervised learning techniques carry a fatal deficiency: once the operational environments are changed, the model may hardly work well or even becomes useless due to the distribution shift between the training data and the new dataset. This paper attempts to address this issue by proposing to apply transfer learning techniques to develop lifetime robust living models for real-world DT systems. This paper presents a framework for developing lifetime data-driven living models. A case study, railway digital twin from our on-going research project along with the preliminary results, will be presented to demonstrate the feasibility and usefulness of the proposed modeling framework for digital twin., 2022 IEEE 18th International Conference on Automation Science and Engineering (CASE), August 20-24, 2022, Mexico City, Mexico

In this paper, flexible pressure sensors based on 3D printing technology are presented. These flexible pressure sensors have different dielectrics and are designed to be a sandwich structure, which consists of a top electrode, flexible dielectric layer and a bottom electrode. The flexible dielectric layer in the middle is the pressure sensitive layer and is designed to be a hollow prism structure. This design is inspired by an origami structure and is intended to achieve excellent mechanical property. Besides that, in order to develop a method of improving flexible dielectric properties, an experiment of the synthetic composite dielectric layer is carried out and three kinds of flexible composite dielectrics are fabricated. Polydimethylsiloxane (PDMS), Barium Titanate (BaTiO3) and Carbon Nanotubes (CNTs) are used to synthesize composite dielectrics. They are PDMS compounded with BaTiO3, PDMS compounded with CNTs, PDMS compounded with both BaTiO3 and CNTs, respectively. The dielectric structure is constructed with a 3D printed mold. The copper foil is patterned to work as electrodes of the sensors. Besides that, a sensor with pure PDMS dielectric is fabricated as a reference. By testing the sensitivity of these sensors, it was found that the sensor with dielectric made of PDMS doping both BaTiO3 and CNTs shows the highest sensitivity, which is about 0.1kPa−1 and 1.3 times of the reference sensor at the pressure below 20kPa. As a result, fabricating composite dielectric in this way can be a potential method to improve the device performance.

Drug discovery is a challenging process with a huge molecular space to be explored and numerous pharmacological properties to be appropriately considered. Among various drug design protocols, fragment-based drug design is an effective way of constraining the search space and better utilizing biologically active compounds. Motivated by fragment-based drug search for a given protein target and the emergence of artificial intelligence (AI) approaches in this field, this work advances the field of in silico drug design by (1) integrating a graph fragmentation-based deep generative model with a deep evolutionary learning process for large-scale multi-objective molecular optimization, and (2) applying protein-ligand binding affinity scores together with other desired physicochemical properties as objectives. Our experiments show that the proposed method can generate novel molecules with improved property values and binding affinities.

The design of a new therapeutic agent is a time-consuming and expensive process. The rise of machine intelligence provides a grand opportunity of expeditiously discovering novel drug candidates through smart search in the vast molecular structural space. In this paper, we propose a new approach called adversarial deep evolutionary learning (ADEL) to search for novel molecules in the latent space of an adversarial generative model and keep improving the latent representation space. In ADEL, a custom-made adversarial autoencoder (AAE) model is developed and trained under a deep evolutionary learning (DEL) process. This involves an initial training of the AAE model, followed by an integration of multi-objective evolutionary optimization in the continuous latent representation space of the AAE rather than the discrete structural space of molecules. By using the AAE, an arbitrary distribution can be provided to the training of AAE such that the latent representation space is set to that distribution. This allows for a starting latent space from which new samples can be produced. Throughout the process of learning, new samples of high quality are generated after each iteration of training and then added back into the full dataset, therefore, allowing for a more comprehensive procedure of understanding the data structure. This combination of evolving data and continuous learning not only enables improvement in the generative model, but the data as well. By comparing ADEL to the previous work in DEL, we see that ADEL can obtain better property distributions. We show that ADEL is able to design high-quality molecular structures which can be further used for virtual and experimental screenings.

Empowering autonomous agents with 3D understanding for daily objects is a grand challenge in robotics applications. When exploring in an unknown environment, existing methods for object pose estimation are still not satisfactory due to the diversity of object shapes. In this paper, we propose a novel framework for category-level object shape and pose estimation from a single RGB-D image. To handle the intra-category variation, we adopt a semantic primitive representation that encodes diverse shapes into a unified latent space, which is the key to establish reliable correspondences between observed point clouds and estimated shapes. Then, by using a SIM(3)-invariant shape descriptor, we gracefully decouple the shape and pose of an object, thus supporting latent shape optimization of target objects in arbitrary poses. Extensive experiments show that the proposed method achieves SOTA pose estimation performance and better generalization in the real-world dataset. Code and video are available at https://zju3dv.github.io/gCasp., Comment: CoRL 2022, 18 pages, 13 figures, typos corrected, references added

The core problem of text-based person retrieval is how to bridge the heterogeneous gap between multi-modal data. Many previous approaches contrive to learning a latent common manifold mapping paradigm following a \textbf{cross-modal distribution consensus prediction (CDCP)} manner. When mapping features from distribution of one certain modality into the common manifold, feature distribution of the opposite modality is completely invisible. That is to say, how to achieve a cross-modal distribution consensus so as to embed and align the multi-modal features in a constructed cross-modal common manifold all depends on the experience of the model itself, instead of the actual situation. With such methods, it is inevitable that the multi-modal data can not be well aligned in the common manifold, which finally leads to a sub-optimal retrieval performance. To overcome this \textbf{CDCP dilemma}, we propose a novel algorithm termed LBUL to learn a Consistent Cross-modal Common Manifold (C$^{3}$M) for text-based person retrieval. The core idea of our method, just as a Chinese saying goes, is to `\textit{san si er hou xing}', namely, to \textbf{Look Before yoU Leap (LBUL)}. The common manifold mapping mechanism of LBUL contains a looking step and a leaping step. Compared to CDCP-based methods, LBUL considers distribution characteristics of both the visual and textual modalities before embedding data from one certain modality into C$^{3}$M to achieve a more solid cross-modal distribution consensus, and hence achieve a superior retrieval accuracy. We evaluate our proposed method on two text-based person retrieval datasets CUHK-PEDES and RSTPReid. Experimental results demonstrate that the proposed LBUL outperforms previous methods and achieves the state-of-the-art performance., Comment: Accepted on ACM MM '22. arXiv admin note: text overlap with arXiv:2209.05773

In natural language processing (NLP), the context of a word or sentence plays an essential role. Contextual information such as the semantic representation of a passage or historical dialogue forms an essential part of a conversation and a precise understanding of the present phrase or sentence. However, the standard attention mechanisms typically generate weights using query and key but ignore context, forming a Bi-Attention framework, despite their great success in modeling sequence alignment. This Bi-Attention mechanism does not explicitly model the interactions between the contexts, queries and keys of target sequences, missing important contextual information and resulting in poor attention performance. Accordingly, a novel and general triple-attention (Tri-Attention) framework expands the standard Bi-Attention mechanism and explicitly interacts query, key, and context by incorporating context as the third dimension in calculating relevance scores. Four variants of Tri-Attention are generated by expanding the two-dimensional vector-based additive, dot-product, scaled dot-product, and bilinear operations in Bi-Attention to the tensor operations for Tri-Attention. Extensive experiments on three NLP tasks demonstrate that Tri-Attention outperforms about 30 state-of-the-art non-attention, standard Bi-Attention, contextual Bi-Attention approaches and pretrained neural language models1.

This paper presents an approach for the topological design of multi-material compliant mechanisms with global stress constraints. The element stacking method and the separable stress interpolation scheme are applied to calculate the element stiffness and element stress of multi-material structures. The output displacement of multi-material compliant mechanisms is maximized under the constraints of the maximum stress and the structural volume of each material. The modified P-norm method is applied to aggregate the local von Mises stress constraints for all the finite elements to a global stress constraint. The sensitivities are calculated by the adjoint method, and the method of moving asymptotes is utilized to update the optimization problem. Several numerical examples are presented to demonstrate the effectiveness of the proposed method. The appearance of the de facto hinges in the optimal mechanisms can be suppressed effectively by using the topology optimization model with global stress constraints, and the stress constraints for each material can be met.

Grasping in cluttered scenes has always been a great challenge for robots, due to the requirement of the ability to well understand the scene and object information. Previous works usually assume that the geometry information of the objects is available, or utilize a step-wise, multi-stage strategy to predict the feasible 6-DoF grasp poses. In this work, we propose to formalize the 6-DoF grasp pose estimation as a simultaneous multi-task learning problem. In a unified framework, we jointly predict the feasible 6-DoF grasp poses, instance semantic segmentation, and collision information. The whole framework is jointly optimized and end-to-end differentiable. Our model is evaluated on large-scale benchmarks as well as the real robot system. On the public dataset, our method outperforms prior state-of-the-art methods by a large margin (+4.08 AP). We also demonstrate the implementation of our model on a real robotic platform and show that the robot can accurately grasp target objects in cluttered scenarios with a high success rate. Project link: https://openbyterobotics.github.io/sscl, Comment: International Conference on Intelligent Robots and Systems (IROS) 2021

The brown planthopper (BPH), Nilaparvata lugens, is an important pest that affects rice (Oryza sativa) production in Asia. The flavone tricin (5,7,4ʹ-trihydroxy-3ʹ,5ʹ-dimethoxy flavone) is a valuable secondary metabolite commonly found in rice plants that can defend rice plants against infestation by BPH. BPH damage can reduce the metabolic level of tricin in rice. Our preliminary transcriptome research results showed that BPH salivary protein 7, NlSP7, is highly responsive to tricin stimuli. However, the function of NlSP7 in mediating the interaction between the rice plant and the BPH is unknown. In this study, we cloned the NlSP7 gene in N. lugens and found that its mRNA level was greater in the presence of high tricin content than low tricin content, regardless of whether the BPHs were fed a rice plant diet or an artificial diet containing 100 mg/L tricin. Knocking down NlSP7 resulted in BPH individuals spending more time in the non-penetration and pathway phase, and less time feeding on the phloem of rice plants. These changes decreased BPH food intake, feeding behavior, and fitness, as well as the tricin content of the rice plants. These findings demonstrate that the salivary protein NlSP7 of BPH functions as an effector for tricin metabolism in rice.

In this paper, we propose a deep evolutionary learning (DEL) process that integrates fragment-based deep generative model and multi-objective evolutionary computation for molecular design. Our approach enables (1) evolutionary operations in the latent space of the generative model, rather than the structural space, to generate novel promising molecular structures for the next evolutionary generation, and (2) generative model fine-tuning using newly generated high-quality samples. Thus, DEL implements a data-model co-evolution concept which improves both sample population and generative model learning. Experiments on two public datasets indicate that sample population obtained by DEL exhibits improved property distributions, and dominates samples generated by multi-objective Bayesian optimization algorithms.

A method for calculating the loss of permanent magnet synchronous motor (PMSM) under Inverter supplied is proposed. It includes calculation of voltage waveform and transient current waveform under multi-mode PWM modulation, and the current waveform is taken as excitation to be input to the finite element simulation model, Then stator core loss, rotor core loss, permanent magnet eddy current loss and stator winding loss considering skin effect are calculated in finite element simulation. The calculated values of the prototype agree well with the measured data and verify the feasibility of the calculation method.

Generating and eliminating adversarial examples has been an intriguing topic in the field of deep learning. While previous research verified that adversarial attacks are often fragile and can be defended via image-level processing, it remains unclear how high-level features are perturbed by such attacks. We investigate this issue from a new perspective, which purely relies on logits, the class scores before softmax, to detect and defend adversarial attacks. Our defender is a two-layer network trained on a mixed set of clean and perturbed logits, with the goal being recovering the original prediction. Upon a wide range of adversarial attacks, our simple approach shows promising results with relatively high accuracy in defense, and the defender can transfer across attackers with similar properties. More importantly, our defender can work in the scenarios that image data are unavailable, and enjoys high interpretability especially at the semantic level.

The modular multilevel converter (MMC) based unified power flow controller (UPFC) is the newest flexible AC transmission system technology at present, which can realize the adjustment of active power and reactive power with voltage regulation, series compensation and shift equalization. With secondary circulation in the MMC bridge arm during UPFC running, a method for calculating secondary circulation current of modular multilevel converters is proposed, which establishes a calculation method of secondary circulation current by simplifying averaging control to be double proportional regulator, based on capacitance volt-ampere characteristic and equivalent switching function. The topological structure of MMC-UPFC is analyzed, and the specific process of the above model is displayed based on understanding the reason of producing secondary circulation current. In order to improve calculation accuracy, correction model of secondary circulation current is established, and the optimal parameters are obtained by using simulated annealing based on least squares. An important conclusion that secondary circulation current is associated with K V1 and K I1 , and it increase with the increase of K V1 and decrease with the increase of K I1, is obtained by analyzing the above model, which provides a theoretical basis for suppressing secondary circulation current and selecting power device for UPFC. Finally PSCAD simulation results show that the method can accurately calculate secondary circulation current, which verifies its validity and feasibility.

Distributed energy generation based on renewable energy sources has grown rapidly in recent decades. The inherent intermittent nature of renewable power brings significant challenges to the grid leading to poor power quality and stability issues. Energy storage is the key technology to address these issues. Among all energy storage technologies, the vanadium redox flow battery (VRB) is regarded as one of the most promising technologies for large scale energy storage applications. When operating the VRB system particularly in conjunction with renewable energy generations, the battery control system plays a critical role in optimising the battery efficiency and maximising the cycle life. This thesis focuses on the developments of the VRB control and management system for grid connected applications. Specifically, it covers four aspects: 1. A dynamic plug flow reactor model is developed for a multi-cell VRB system, which includes the self-discharging reactions, ohmic voltage losses and over-potential losses. Simulation studies were conducted to investigate the effects of flow rate, cut-off state of charge (SOC) limit and input current. 2. A model based optimisation approach is proposed to obtain the optimal charging current and flow rate for the VRB system. The proposed charging regime minimises the total energy consumption of the overall system and simultaneously satisfies the system physical constraints. Two cases studies are presented to charge the battery under different time constraints. 3. A gain scheduling approach is proposed to control the flow rate for the VRB system. Due to the lack of reliable flow rate measurements, the flow rate is regulated by manipulating the pump speed such that the difference between the open circuit voltages before and after the cell stack follows a desired trajectory. 4. To study the feasibility of using the VRB for power quality control applications, experiments were conducted to investigate the dynamic responses of the VRB to load changes over a range of frequencies, through experimental studies on a laboratory scale VRB cell. Experiments were carried out under different operating conditions to examine the effects of system SOC, discharging current and temperature. An equivalent circuit model is developed to describe the battery transient behaviour.

In this Letter, flexible capacitive pressure sensors are developed. Compared to the current mainstream design, the structure is easier to get a large deformation in the dielectric layer and its deformation is more controllable. Since the structure is symmetrical, it deforms uniformly and its Poisson's ratio is near zero when pressure is applied. Besides that, by moderating the structure dimension, sensors with different sensitivity and detection pressure range can be achieved. The fabrication process of these flexible sensors is mainly based on the 3D printing technology, with which the moulds of the sensor are made. The mixture of polydimethylsiloxane and hardening solution is dripped into the printed mould to form the dielectric layer. When it is completely solidified, the dielectric layer is demoulded from the mould. Four moulds with different inner sizes are printed and then four sensors are fabricated and tested. The measurement results show that the capacitance of the sensors increases when the applied pressure increases. The sensor with thinner supporting structure in the dielectric layer has higher sensitivity. The sensor with a thicker supporting structure in the dielectric layer has a larger pressure detection range. All the sensors have low hysteresis error and excellent repeatable performance.

Background and aim Tobacco–rice rotation is a useful traditional practice for controlling crop diseases and pests. However, its effect on rice pests has not been studied. We aimed to clarify the ecological effect of tobacco–rice rotation on two destructive rice pests Nilaparvata lugens (Stal) and Sogatella furcifera (Horvath).

Microwave attenuation materials with high thermal conductivity are required for developing high power microwave technology. In this paper, boron-doped diamond films with different doping concentrations were prepared by microwave plasma chemical vapor deposition. Complex permittivity of the samples was measured in the K-band by using the transmission/reflection method. It was found that the complex permittivity of the diamond films increased with an increase in the boron doping concentration and the diamond films were transformed from a microwave transparent material into a microwave absorbing material. In addition, although thermal conductivity decreased with increasing boron concentration, it remained at a fairly high level. Therefore, boron-doped diamond films could be developed into a microwave attenuation material with extremely high thermal conductivity. Mechanism analysis revealed that the increase in the real part of permittivity mainly resulted from the hopping polarization of bound charges, while the increase in the imaginary part was due to both hopping polarization and valence band conduction. It was found that with the increase in the boron doping concentration, the proportion of the dielectric loss of the first mechanism increased.

In this paper, some capacitive pressure sensor related work is presented. The theoretical model of the touch mode elastic membrane was firstly derived based on the energy method in the theoretical part. Three capacitive pressure sensors with different sizes are designed and fabricated. The core part of the sensors is an array of capacitors. Each capacitor has an elastic membrane. When pressure is applied, the elastic membranes bend down and the distance between the top and bottom electrodes decreases. Hence, the capacitance of the sensors increases. The fabrication process is based on both CMOS and post-micro-electro-mechanical systems (MEMS) technology. The sensors are characterized under atmospheric pressure ranging from 10 kPa (kilopascal) to 100 kPa. They work well except for having a high nonlinear error, which is mainly caused by the limitation of the sensing mechanism. Inspired by the testing results of these sensors, an attempt is made to improve the linearity of capacitive pressure sensors. This method is based on the complementary sensitivity characterization of touch mode and non-touch mode capacitors. By adjusting the sizes and numbers of the touch mode capacitors and non-touch mode capacitors in the sensor, the nonlinear error of capacitive pressure sensors can be reduced and even eliminated. To verify the feasibility of this method, a structure is designed and simulated with FEM software. The simulation results show that this method is effective at improving the linearity of capacitive sensors. The nonlinear error is remarkably reduced to 2.5%, which is about 28% for the contact mode capacitor and 13.6% for the non-touch mode capacitor.

Payments for Ecosystem Services(PES) are thought to be an effective method to solve eco-environmental problems,and it′s widely adopted to promote ecological conservation all around the world.During the application of PES,Establishing the standard of compensation is crucial,because unreasonable standard may not compensate for the losses of stakeholders,which may dampen participants′ enthusiasm and influence the effects of PES. Sanjiangyuan Natural Reserve,which is one of the most important natural reserves of China,plays key role to the ecological security of lower watershed in China and Southeast Asia.Before setting up this reserve,there are large numbers of herdsmen living in this area,and conservation projects in reserve,including "Return Husbandry to Grassland","Return Grain for Green","Ecological Migration" and so on,lead to a series of losses of local residents.So giving these migrations compensation through PES is important to the development of reserve and local social harmony. During this research,we take two migration villages of Sanjiangyuan natural reserve(GuoLuo village and HeYuan village) as example to calculate the standard of compensation.Basing on the practical situation of research area,and considering the common problems of Chinese PES projects,we adopt 3 types of approaches("opportunity cost of livestock","opportunity cost of pasture" and "difference of regional development").Because the preferences of these approaches are different,the established standards of compensation have slight difference. The result of "opportunity cost of livestock" is 13.9 thousand Yuan per household per year.Livestock is the main source of herdsmen′s revenue,and the ecological migration project let them lose this part."Opportunity cost of livestock" takes livestock as proxy when calculating compensation standard,and can compensate for their direct economic losses.So this approach is strongly correlated to the benefits of herdsmen and can maximize the compensation. The result of "opportunity cost of pasture" is 10.3 thousand Yuan per household per year.Herdsmen can rent out their pasture to others to get rent,and the ecological migration project also let them lose this part of income."Opportunity cost of pasture" takes pasture as proxy,and compensate for the losses of rent.This approach considers pasture as commodity and reflects the relationship between supply and demand,and this approach can also simplify the calculating process.But this approach doesn′t consider natural heterogeneity,and all the pastures are calculated equally. The result of "difference of regional development" is 11 thousand Yuan per household per year.There is great regional economic development difference in our research area,and this actuality should not be ignored during the PES project,especially for ecological migration."Difference of regional development" can consider this difference,but the standard is still too low,comparing to the other two approaches.And neither herdsmen′ losses nor ecosystem services are related to this approach. Our results suggests that PES may provide limited benefits for the poverty reduction,but this research can still provide scientific foundation of establishing the compensation standard for Sanjiangyuan natural reserve,and case studying in other area can also borrow idea from our research.

Ecosystems provide human multiple types of ecosystem services in the aspects of natural resources and living environment,and all these services are crucial foundation of sustainable social and economic development.However,between the understanding of ecosystem services and practice of management,there exist great challenges which contain: quantifying ecosystem services,relationships among multiple services,multi-scale correlation of services,combination of ecosystem services and policy design.In response to these challenges,researches of ecosystem services management in recent years have focused on fields as follow: quantitative estimate of ecosystem services,relationship between ecosystem services and human well-being,trade-offs among multiple ecosystem services,conservation planning of ecosystem services and mechanisms of payments for ecosystem services.With the goal to promote ecosystem services management practices,we should: further strengthen the theoretical research about provision of ecosystem services,develop more methods to display results of services research,carry out more interdisciplinary studies with sociology,economics,demography,search for practicable and reasonable approach to incorporating ecosystem services into decision-making.

Ecosystem services are the goods and services provided by ecosystems to people.Ecosystem services are important basis of human existence and strongly related to human well-being,and the sustainable provision of multiple services is essential for the sustenance of human societies.There are a series of factors driving ecosystem services,one of which is the change of land use pattern,and it is drawing more and more attention of ecologists. Miyun reservoir,which is the only surface water source for domestic water in Beijing,plays the key role for the ecological security of the capital.Miyun reservoir watershed is a vital water source region of Beijing city,and it has many important ecosystem services.The water yield and water quality of reservoir strongly depend on Miyun reservoir watershed′s condition,therefore,the conservation and improvement of ecosystem services in Miyun reservoir watershed has vital significance for local and Beijing′s ecological security.However,how the land use change influenced ecosystem services during the past decades is not clear yet,which may block the comprehensive development of watershed. So we take Miyun reservoir watershed as example,analyze the change of land use and ecosystem services between 1990 and 2009.We set up a two-grade classify system of land use and interpreted the remote sensing images of 1990,2000 and 2009 respectively,and then apply Integrated Valuation of Ecosystem Services and Tradeoffs(InVEST model),which is a space explicit model and based on land use,to map the ecosystem services,and analyze the influence of Land Use to ecosystem services during study period.The sub models we apply in this research contain "Water Yield","Sediment Retention" and "Water Purification",and before running these sub models,we verify them with observed data. The results of image interpretation show that farmland,forest and grassland are main type of land use of watershed,and these areas account for more than 95% of the whole watershed.There are sharp changes of land use during study period,and conversion of land use types distributed more widely after 2000.The general trend in study period is farmland and grassland decreased by 30%,48% respectively,when forest increased by 30%,and another change cannot be neglected is residential area expanded by more than 280% during the past 20 years. The changes of land use have affect on the pattern of ecosystem services and there are trade-offs among different ecosystem services in question.During the study period,the decreasing of farmland and increasing of forest and grassland improved the ability of sediment retention,and this service increased by 46% in the past 20 years.But forest expansion caused more evapotranspiration,which reduced water yield of watershed by 3% at the same time.Because the nutrient loading of residential area is much higher than other types of land use,the large-scale expansion of residential area decreased water purification service sharply after 2000 although the expansion of forest and decrease of farmland can improve this service to a certain degree,and this service declined by 55% in the past 20 years.With the expansion of forest,carbon sequestration service increased about 19%.So with the goal to get overall improvement of ecosystem services in watershed,we suggest that manages should strengthen the control of forest and residential areas. Understanding the relationship between land use and ecosystem services,the department of decision-making can control and reduce the trade-off between ecosystem services and achieve the sustainable development through appropriate land use plan.Although the approach and process of evaluation still need more improvement,our research can provide scientific foundation for decision-making in Miyun reservoir watershed.

The meaning of parsimony is twofold in machine learning: either the structure or (and) the parameter of a model can be sparse. Sparse models have many strengths. First, sparsity is an important regularization principle to reduce model complexity and therefore avoid overfitting. Second, in many fields, for example bioinformatics, many high-dimensional data may be generated by a very few number of hidden factors, thus it is more reasonable to use a proper sparse model than a dense model. Third, a sparse model is often easy to interpret. In this dissertation, we investigate the sparse machine learning models and their applications in high-dimensional biological data analysis. We focus our research on five types of sparse models as follows. First, sparse representation is a parsimonious principle that a sample can be approximated by a sparse linear combination of basis vectors. We explore existing sparse representation models and propose our own sparse representation methods for high dimensional biological data analysis. We derive different sparse representation models from a Bayesian perspective. Two generic dictionary learning frameworks are proposed. Also, kernel and supervised dictionary learning approaches are devised. Furthermore, we propose fast active-set and decomposition methods for the optimization of sparse coding models. Second, gene-sample-time data are promising in clinical study, but challenging in computation. We propose sparse tensor decomposition methods and kernel methods for the dimensionality reduction and classification of such data. As the extensions of matrix factorization, tensor decomposition techniques can reduce the dimensionality of the gene-sample-time data dramatically, and the kernel methods can run very efficiently on such data. Third, we explore two sparse regularized linear models for multi-class problems in bioinformatics. Our first method is called the nearest-border classification technique for data with many classes. Our second method is a hierarchical model. It can simultaneously select features and classify samples. Our experiment, on breast tumor subtyping, shows that this model outperforms the one-versus-all strategy in some cases. Fourth, we propose to use spectral clustering approaches for clustering microarray time-series data. The approaches are based on two transformations that have been recently introduced, especially for gene expression time-series data, namely, alignment-based and variation-based transformations. Both transformations have been devised in order to take into account temporal relationships in the data, and have been shown to increase the ability of a clustering method in detecting co-expressed genes. We investigate the performances of these transformations methods, when combined with spectral clustering on two microarray time-series datasets, and discuss their strengths and weaknesses. Our experiments on two well known real-life datasets show the superiority of the alignment-based over the variation-based transformation for finding meaningful groups of co-expressed genes. Fifth, we propose the max-min high-order dynamic Bayesian network (MMHO-DBN) learning algorithm, in order to reconstruct time-delayed gene regulatory networks. Due to the small sample size of the training data and the power-low nature of gene regulatory networks, the structure of the network is restricted by sparsity. We also apply the qualitative probabilistic networks (QPNs) to interpret the interactions learned. Our experiments on both synthetic and real gene expression time-series data show that, MMHO-DBN can obtain better precision than some existing methods, and perform very fast. The QPN analysis can accurately predict types of influences and synergies. Additionally, since many high dimensional biological data are subject to missing values, we survey various strategies for learning models from incomplete data. We extend the existing imputation methods, originally for two-way data, to methods for gene-sample-time data. We also propose a pair-wise weighting method for computing kernel matrices from incomplete data. Computational evaluations show that both approaches work very robustly.

In order to solve the elastic wave equation in heterogeneous media with arbitrary high order accuracy in space on unstructured meshes, a nodal Discontinuous Galerkin Finite Element Method (DG-FEM) is presented, which combines the geometrical flexibility of the Finite Element Method and strongly nonlinear wave simulation capability of the Finite Volume Me-thod. The equations of nonlinear elastodynamics have been written in a conservative form in order to facilitate the numerical implementation and introduce different kinds of elastic non-linearities, such as the classical nonlinearities and non-classical hysteretic nonlinearities. In the calculation of DG-FEM scheme, different kinds of boundary conditions and numerical fluxes have been discussed. The numerical simulations of linear elastic wave propagation and plane wave nonlinear propagation demonstrated the developed DG-FEM scheme has an excellent precision and performance in numerical application.

In this thesis we propose the development of an innovative micro-damage imaging system based on a combination of Nonlinear Elastic Wave Spectroscopy techniques and “chaotic cavity transducer” concept. It consists of a combination of a PZT ceramic glued to a cavity of chaotic shape with the time reversal principle. The feasibility and capabilities of these new ideas is explored by numerical simulations, and optimal operational parameters for experimental implementation are suggested based on the modelling support. A large part of the research work conducted in this thesis is concentrated on the development of numerical simulation tools to help the improvement of such nonlinear imaging methods. A nodal Discontinuous Galerkin Finite Element Method (DG-FEM) scheme is extended to nonlinear elasto-dynamic including source terms. A Perfectly Matched Layer absorbing boundary condition well adapted to the DG-FEM scheme, called Nearly Perfectly Matched Layer (NPML), is also developed. In the case of orthotropic material as stability problems appear, a mixture of NPML and sponge layer, with a controllable ratio of these two kinds of absorbing layers, is introduced. The experimental validation of “chaotic cavity transducer” to focalize in reverberant and non-reverberant solid media with only one source is made. Classical time reversal, inverse filter and 1 Bit time reversal process are discussed and compared. The experimental demonstration of the use of a “chaotic cavity transducer”, in combination with the pulse inversion and 1-bit methods, to obtain an image of localized nonlinearity is made. This opens the possibility for high resolution imaging of nonlinear defects; Dans cette thèse nous proposons de développer un système d’imagerie ultrasonore innovante de micro- défauts basé sur l’utilisation conjointe de techniques d’acoustique non linéaire et du concept de "transducteur à cavité chaotique". Ce transducteur correspond à la combinaison d’une céramique piézoélectrique collée sur une cavité de forme chaotique et du principe de retournement temporel. La faisabilité et les performances de ce nouveau système sont explorées par des simulations numériques. Des paramètres optimaux d’utilisation pour une implémentation expérimentale sont proposés. Une grande partie des travaux menés dans le cadre de cette thèse se concentre sur le développement d’outils numériques permettant l’amélioration de telles techniques d’imagerie. Un schéma d’éléments finis de type Galerkin Discontinu (GD) est étendu à l’élastodynamique non linéaire. Un type de zone absorbante parfaitement adaptée, appelée "Nearly Perfectly Matched Layer" (NPML) a aussi été développé. Dans le cas de matériaux orthotropes, comme des problèmes de stabilité apparaissent, un mélange de NPML et de zone atténuante, dont on contrôle la proportion respective, est introduit afin de stabiliser les NPML. Une validation expérimentale du concept de "transducteur à cavité chaotique" pour la focalisation dans un milieu solide, réverbérant ou non, en utilisant une seule source est réalisée. Les méthodes de retournement temporel et de filtre inverse sont présentées et comparées. La démonstration expérimentale qu’un "transducteur à cavité chaotique" peut être utilisé conjointement avec les méthodes d’inversion d’impulsion afin de réaliser une image de non linéarités localisées est présentée

In this thesis we propose the development of an innovative micro-damage imaging system based on a combination of Nonlinear Elastic Wave Spectroscopy techniques and “chaotic cavity transducer” concept. It consists of a combination of a PZT ceramic glued to a cavity of chaotic shape with the time reversal principle. The feasibility and capabilities of these new ideas is explored by numerical simulations, and optimal operational parameters for experimental implementation are suggested based on the modelling support. A large part of the research work conducted in this thesis is concentrated on the development of numerical simulation tools to help the improvement of such nonlinear imaging methods. A nodal Discontinuous Galerkin Finite Element Method (DG-FEM) scheme is extended to nonlinear elasto-dynamic including source terms. A Perfectly Matched Layer absorbing boundary condition well adapted to the DG-FEM scheme, called Nearly Perfectly Matched Layer (NPML), is also developed. In the case of orthotropic material as stability problems appear, a mixture of NPML and sponge layer, with a controllable ratio of these two kinds of absorbing layers, is introduced. The experimental validation of “chaotic cavity transducer” to focalize in reverberant and non-reverberant solid media with only one source is made. Classical time reversal, inverse filter and 1 Bit time reversal process are discussed and compared. The experimental demonstration of the use of a “chaotic cavity transducer”, in combination with the pulse inversion and 1-bit methods, to obtain an image of localized nonlinearity is made. This opens the possibility for high resolution imaging of nonlinear defects; Dans cette thèse nous proposons de développer un système d’imagerie ultrasonore innovante de micro- défauts basé sur l’utilisation conjointe de techniques d’acoustique non linéaire et du concept de "transducteur à cavité chaotique". Ce transducteur correspond à la combinaison d’une céramique piézoélectrique collée sur une cavité de forme chaotique et du principe de retournement temporel. La faisabilité et les performances de ce nouveau système sont explorées par des simulations numériques. Des paramètres optimaux d’utilisation pour une implémentation expérimentale sont proposés. Une grande partie des travaux menés dans le cadre de cette thèse se concentre sur le développement d’outils numériques permettant l’amélioration de telles techniques d’imagerie. Un schéma d’éléments finis de type Galerkin Discontinu (GD) est étendu à l’élastodynamique non linéaire. Un type de zone absorbante parfaitement adaptée, appelée "Nearly Perfectly Matched Layer" (NPML) a aussi été développé. Dans le cas de matériaux orthotropes, comme des problèmes de stabilité apparaissent, un mélange de NPML et de zone atténuante, dont on contrôle la proportion respective, est introduit afin de stabiliser les NPML. Une validation expérimentale du concept de "transducteur à cavité chaotique" pour la focalisation dans un milieu solide, réverbérant ou non, en utilisant une seule source est réalisée. Les méthodes de retournement temporel et de filtre inverse sont présentées et comparées. La démonstration expérimentale qu’un "transducteur à cavité chaotique" peut être utilisé conjointement avec les méthodes d’inversion d’impulsion afin de réaliser une image de non linéarités localisées est présentée

The transmission of optical signals under the effects of cross-phase modulation (XPM) and self-phase modulation (SPM) together in pump-probe wave structure is investigated in intensity modulation-direct detection wavelength division multiplexing (IM-DD WDM) systems. A simple expression for frequency domain IM of probe channel is derived. The analytical method to determine the temporal shape of probe wave when arbitrary wave is inputted in pump channel is confirmed. The impact of channel separation and transmission rate to XPM is assessed. We find that the SPM in pump channel increase temporal intensity modulation index reflected XPM to a certain extent. The analytical result agrees well with numerical simulation.

This paper applies a pattern recognition method for localisation of the PD site in the oil-filled transformer by acoustic measurements. By dividing the transformer into N numbers of 20/spl times/20/spl times/20 mm/sup 3/ sub-modules, the PD site is localised by comparing the spatial distances of each sub-module to the acoustic sensors. The numerical cases indicate that the proposed method is general and practical while avoiding some drawbacks in current algorithms. Moreover, the method improves the accuracy of localisation substantially when compared with the conventional localization method.

The effect of biofilms developed in the aquariums on settlement of plantigrades of the mussel Mytilus coruscus was examined in the laboratory.Various aspects of biofilm biology were investigated including age,dry weight,chlorophyll a(Chl.a)content,bacterial and diatom densities.Bacterial community in biofilms was analyzed by the DGGE fingerprint technique.The correlation between the inducing activity of the biofilms and dry weight,bacterial and diatom densities,Chl.a content was also investigated.With the increase of biofilm age,dry weight,bacterial and diatom densities increased significantly and the maximum values were,respectively,0.87 mg/cm2,1.5×107/cm2 and 1.0×106/cm2 after 28 d.Furthermore,above three parameters were correlated with biofilm age.Chl.a content reached the maximum of 2.22 μg/cm2 after 14 d and no correlation between Chl.a content and biofilm age was observed.Plantigrades of M.coruscus preferred to settle on the biofilmed surfaces rather than the surfaces unfilmed.Overall the settlement of plantigrades on biofilms that formed in the aquariums increased with biofilm age,and the maximum of 76% settlement was obtained after 28 d.The enhanced settlement correlated to dry weight,bacterial density and diatom density,and their corresponding correlation coefficients were 0.717,0.711 and 0.754.In contrast,there was no correlation between inducing activity of biofilms and Chl.a content.Moreover,Shannon diversity index of bacteria community in DGGE profiles increased up to 28 d following immersion.The shift in bacterial community composition over time may be responsible for the higher settlement of plantigrades of M.coruscus on the older biofilms,indicating that bacterial community play an important role in the settlement of plantigrades in this species.