Robots that use vision for localization need to handle environments that are subject to seasonal and structural change, and operate under changing lighting and weather conditions. We present a framework for lifelong localization and mapping designed to provide robust and metrically accurate online localization in these kinds of changing environments. Our system iterates between offline map building, map summary, and online localization. The offline mapping fuses data from multiple visually varied datasets, thus dealing with changing environments by incorporating new information. Before passing these data to the online localization system, the map is summarized, selecting only the landmarks that are deemed useful for localization. This Summary Map enables online localization that is accurate and robust to the variation of visual information in natural environments while still being computationally efficient. We present a number of summary policies for selecting useful features for localization from the multisession map, and we explore the tradeoff between localization performance and computational complexity. The system is evaluated on 77 recordings, with a total length of 30 kilometers, collected outdoors over 16 months. These datasets cover all seasons, various times of day, and changing weather such as sunshine, rain, fog, and snow. We show that it is possible to build consistent maps that span data collected over an entire year, and cover day‐to‐night transitions. Simple statistics computed on landmark observations are enough to produce a Summary Map that enables robust and accurate localization over a wide range of seasonal, lighting, and weather conditions. 相似文献
Free Material Optimization (FMO) is a powerful approach for conceptual optimal design of composite structures. The design variable in FMO is the entire elastic material tensor which is allowed to vary almost freely over the design domain. The imposed requirements on the tensor are that it is symmetric and positive semidefinite. Most of today’s studies on FMO focus on models for two- and three-dimensional structures. The objective of this article is to extend existing FMO models and methods to laminated plate and shell structures, which are used in many engineering applications. In FMO, the resulting optimization problem is generally a non convex semidefinite program with many matrix inequalities which requires special-purpose optimization methods. The FMO problems are efficiently solved by a primal-dual interior point method developed and implemented by the authors. The quality of the proposed FMO models and the method are supported by several large-scale numerical experiments. 相似文献
Testing model transformations poses several challenges, among them the automatic generation of appropriate input test models and the specification of oracle functions. Most approaches for the generation of input models ensure a certain coverage of the source meta-model or the transformation implementation code, whereas oracle functions are frequently defined using query or graph languages. However, these two tasks are usually performed independently regardless of their common purpose, and sometimes, there is a gap between the properties exhibited by the generated input models and those considered by the transformations. Recently, we proposed a formal specification language for the declarative formulation of transformation properties (by means of invariants, pre-, and postconditions) from which we generated partial oracle functions used for transformation testing. Here, we extend the usage of our specification language for the automated generation of input test models by SAT solving. The testing process becomes more intentional because the generated models ensure a certain coverage of the transformation requirements. Moreover, we use the same specification to consistently derive both the input test models and the oracle functions. A set of experiments is presented, aimed at measuring the efficacy of our technique. 相似文献
In light of the growing relevance of customer-oriented business strategies IT investments in the field of Customer Relationship Management have increased considerably. However, firms often could not realize sufficient returns on these IT investments. One major reason for this failure seems to be the lack of appropriate approaches to determine the economic impact of such investments ex ante. Therefore, we develop an economic model to determine the optimal level of Customer Relationship Management IT investments. Using this approach, firms can evaluate, to what extent investments in Customer Relationship Management IT are reasonable. One major result is that in most cases the “all or nothing strategy” pursued by many firms does not lead to the optimal level of investments. To illustrate the practical utility and applicability of the approach, we provide a real world example of a German financial services provider. 相似文献
Several real problems involve the classification of data into categories or classes. Given a data set containing data whose classes are known, Machine Learning algorithms can be employed for the induction of a classifier able to predict the class of new data from the same domain, performing the desired discrimination. Some learning techniques are originally conceived for the solution of problems with only two classes, also named binary classification problems. However, many problems require the discrimination of examples into more than two categories or classes. This paper presents a survey on the main strategies for the generalization of binary classifiers to problems with more than two classes, known as multiclass classification problems. The focus is on strategies that decompose the original multiclass problem into multiple binary subtasks, whose outputs are combined to obtain the final prediction. 相似文献
We present a mathematical analysis of the effects of Hebbian learning in random recurrent neural networks, with a generic Hebbian learning rule, including passive forgetting and different timescales, for neuronal activity and learning dynamics. Previous numerical work has reported that Hebbian learning drives the system from chaos to a steady state through a sequence of bifurcations. Here, we interpret these results mathematically and show that these effects, involving a complex coupling between neuronal dynamics and synaptic graph structure, can be analyzed using Jacobian matrices, which introduce both a structural and a dynamical point of view on neural network evolution. Furthermore, we show that sensitivity to a learned pattern is maximal when the largest Lyapunov exponent is close to 0. We discuss how neural networks may take advantage of this regime of high functional interest. 相似文献
Automatic classification of high-resolution mass spectrometry data has increasing potential to support physicians in diagnosis of diseases like cancer. The proteomic data exhibit variations among different disease states. A precise and reliable classification of mass spectra is essential for a successful diagnosis and treatment. The underlying process to obtain such reliable classification results is a crucial point. In this paper such a method is explained and a corresponding semi automatic parameterization procedure is derived. Thereby a simple straightforward classification procedure to assign mass spectra to a particular disease state is derived. The method is based on an initial preprocessing stage of the whole set of spectra followed by the bi-orthogonal discrete wavelet transform (DWT) for feature extraction. The approximation coefficients calculated from the scaling function exhibit a high peak pattern matching property and feature a denoising of the spectrum. The discriminating coefficients, selected by the Kolmogorov–Smirnov test are finally used as features for training and testing a support vector machine with both a linear and a radial basis kernel. For comparison the peak areas obtained with the it ClinProt-System1 [33] were analyzed using the same support vector machines. The introduced approach was evaluated on clinical MALDI-MS data sets with two classes each originating from cancer studies. The cross validated error rates using the wavelet coefficients where better than those obtained from the peak areas2. 相似文献
The muscular dystrophies are a large and heterogeneous group of neuromuscular disorders that can be classified according to the mode of inheritance, the clinical phenotype and the molecular defect. To better understand the pathological mechanisms of dysferlin myopathy we compared the protein-expression pattern in the muscle biopsies of six patients with this disease with six patients with limb girdle muscular dystrophy 2A, five with facioscapulohumeral dystrophy and six normal control subjects. To investigate differences in the expression levels of skeletal muscle proteins we used 2-DE and MS. Western blot or immunohistochemistry confirmed relevant results. The study showed specific increase expression of proteins involved in fast-to-slow fiber type conversion (ankyrin repeat protein 2), type I predominance (phosphorylated forms of slow troponin T), sarcomere stabilization (actinin-associated LIM protein), protein ubiquitination (TRIM 72) and skeletal muscle differentiation (Rho-GDP-dissociation inhibitor ly-GDI) in dysferlin myopathy. As anticipated, we also found differential expression of proteins common to all the muscular dystrophies studied. This comparative proteomic analysis suggests that in dysferlin myopathy (i) the type I fiber predominance is an active process of fiber type conversion rather than a selective loss of type II fibers and (ii) the dysregulation of proteins involved in muscle differentiation further confirms the role of dysferlin in this process. 相似文献
IT systems pervade our society more and more, and we become heavily dependent on them. At the same time, these systems are increasingly targeted in cyberattacks, making us vulnerable. Enterprise and cybersecurity responsibles face the problem of defining techniques that raise the level of security. They need to decide which mechanism provides the most efficient defense with limited resources. Basically, the risks need to be assessed to determine the best cost-to-benefit ratio. One way to achieve this is through threat modeling; however, threat modeling is not commonly used in the enterprise IT risk domain. Furthermore, the existing threat modeling methods have shortcomings. This paper introduces a metamodel-based approach named Yet Another Cybersecurity Risk Assessment Framework (Yacraf). Yacraf aims to enable comprehensive risk assessment for organizations with more decision support. The paper includes a risk calculation formalization and also an example showing how an organization can use and benefit from Yacraf.