Army Command College

Multitask joint learning technology continues gaining more attention as a paradigm shift and has shown promising performance in many applications. Depth estimation and semantic understanding from monocular images emerge as a challenging problem in computer vision. While the other joint learning frameworks establish the relationship between the semantics and depth from stereo pairs, the lack of learning camera motion renders the frameworks that fail to model the geometric structure of the image scene. We make a further step in this article by proposing a multitask learning method, namely DPSNet, which can jointly perform depth and camera pose estimation and semantic scene segmentation. Our core idea for depth and camera pose prediction is that we present the rigid semantic consistency loss to overcome the limitation of moving pixels from image reconstruction technology and further infer the segmentation of moving instances based on them. In addition, our proposed model performs semantic segmentation by reasoning the geometric correspondences between the pixel semantic outputs and the semantic labels at multiscale resolutions. Experiments on open-source datasets and a video dataset captured on a micro-smart car show the effectiveness of each component of DPSNet, and DPSNet achieves state-of-the-art results in all three tasks compared with the best popular methods. All our models and code are available at https://github.com/jn-z/DPSNet: Multitask Learning Using Geometry Reasoning for Scene Depth and semantics.

Decision-tree algorithm provides one of the most popular methodologies for symbolic knowledge acquisition. The resulting knowledge, a symbolic decision tree along with a simple inference mechanism, has been praised for comprehensibility. The most comprehensible decision trees have been designed for perfect symbolic data. Over the years, additional methodologies have been investigated and proposed to deal with continuous or multi-valued data, and with missing or noisy features. Recently, with the growing popularity of fuzzy representation, some researchers have proposed to utilize fuzzy representation in decision trees to deal with similar situations. This paper presents a survey of current methods for Fuzzy Decision Tree (FDT) designment and the various existing issues. After considering potential advantages of FDT classifiers over traditional decision tree classifiers, we discuss the subjects of FDT including attribute selection criteria, inference for decision assignment and stopping criteria. To be best of our knowledge, this is the first overview of fuzzy decision tree classifier.

In cognitive radio networks (CRNs), a wireless user with primary access right on a channel (called primary user) has prioritized access to the channel and the user with secondary access right (called secondary user) can use the channel when the primary user is idle. Spectrum auction has emerged as a promising approach to address the access allocation problem in CRNs. A significant challenge in designing such auction is providing truthfulness to avoid market manipulation. In most previous work, the primary access rights on channels are pre-determined before the auction and bidders can only compete for the secondary access rights. However, a user's requirement on spectrum access rights relies on their QoS demands. Therefore, it is much desirable to allocate spectrum access rights on the basis of QoS demands as well as to exploit the resulting spatial spectrum reuse opportunities. To solve this problem, we propose TRUMP, a truthful spectrum auction mechanism, by taking into consideration both QoS demands and spectrum spatial reuse, which can drastically improve spectrum utilization. The theoretical analysis proves that TRUMP achieves truthfulness and individual rationality with polynomial-time complexity. Our extensive simulation results show that our proposals outperform previous work in terms of both social welfare and spectrum utilization.

Auction is a promising approach for dynamic spectrum access in cognitive radio networks. Existing auction mechanisms are mainly strategy-proof to stimulate bidders to reveal their valuations of spectrum truthfully. However, they can suffer significantly from a new cheating pattern, named false-name bids, where a bidder can manipulate the auction by submitting bids under multiple fictitious names. We show such false-name bid cheating is easy to make but difficult to detect in dynamic spectrum auctions. To address this issue, we propose ALETHEIA, a novel flexible, false-name-proof auction framework for large-scale dynamic spectrum access. ALETHEIA not only guarantees strategy-proofness but also resists false-name bids. Moreover, ALETHEIA enables spectrum reuse across a large number of bidders, to improve spectrum utilization. Following that, we extend ALETHEIA to its general version that supports more practical and flexible auction, where bidders accept the spectrum allocation under their partial satisfactions. Theoretical analysis and simulation results show that ALETHEIA achieves both high spectrum redistribution efficiency and auction efficiency.

Data streams mining has become a novel research topic of growing interest in knowledge discovery. Because of the high speed and huge size of data set in data streams, the traditional classification technologies are no longer applicable. In recent years a great deal of research has been done on this problem, most intends to efficiently solve the data streams mining problem with concept drift. This paper presents the state-of-the-art in this field with growing vitality and introduces the methods for detecting concept drift in data stream, then gives a critical summary of existing approaches to the problem, including Stagger, FLORA, MetaL(B), MetaL(IB), CD3, CD4, CD5, OLIN, CVFDT and different ensemble classifiers. At last, this paper explores the challenges and future work in this field.

To estimate postgrouting rock mass strength growth is important for engineering design. In this paper, using self‐developed indoor pressure‐grouting devices, 19 groups of test cubic blocks were made of the different water cement ratio grouting into the broken rock of three kinds of particle sizes. The shear strength parameters of each group under different conditions were tested. Then this paper presents a quantitative calculation method for predicting the strength growth of grouted broken rock. Relational equations were developed to investigate the relationship between the growth rates of uniaxial compressive strength (UCS), absolute value of uniaxial tensile strength (AUTS), internal friction angle, and cohesion for post‐ to pregrouting broken rock based on Mohr‐Coulomb strength criterion. From previous test data, the empirical equation between the growth rate of UCS and the ratio of the initial rock mass UCS to the grout concretion UCS has been determined. The equations of the growth rates of the internal friction coefficient and UCS for grouting broken rock with rock mass rating (RMR) and its increment have been established. The calculated results are consistent with the experimental results. These observations are important for engineered design of grouting reinforcement for broken rock mass.

In this paper, we reveal a second-order phase transition for the Dicke model with the dipole–dipole interaction between the atoms. By means of Holstein–Primakoff transformation, the energy spectra for both the normal and superradiant phases are explicitly obtained and therefore the scaling behaviour near the critical transition point can be given clearly. It is also shown that the dipole–dipole interactions between the atoms deeply affect the critical transition point, the ground-state and excitation energies and the scaled atomic inversion in the ground state.

Community detection based on label propagation algorithm (LPA) has attracted widespread concern because of its high efficiency. But it is difficult to guarantee the accuracy of community detection as the label spreading is random in the algorithm. In response to the problem, an improved LPA based on random walk (RWLPA) is proposed in this paper. Firstly, a matrix measuring similarity among various nodes in the network is obtained through calculation. Secondly, during the process of label propagation, when a node has more than a neighbor label with the highest frequency, not the label of a random neighbor but the label of the neighbor with the highest similarity will be chosen to update. It can avoid label propagating randomly among communities. Finally, we test LPA and the improved LPA in benchmark networks and real-world networks. The results show that the quality of communities discovered by the improved algorithm is improved compared with the traditional algorithm.

In order to overcome the subjective randomness in traditional fuzzy comprehensive evaluation, the weight value is modified based on the differential degree of evaluation indices, a new model for comprehensive evaluation-entropy weight fuzzy comprehensive evaluation model is established by using the concept of entropy in information engineering science. Then the model is applied to the optimization of engineering machinery. Calculation and result analysis were carried out through a loader case study. Results show that fuzzy comprehensive evaluation based on entropy weight method suits with the optimization of engineering machinery in which factors interact on each other. This method ensures the validity of optimization result by combining the external and subjective evaluation together, so it has upper applied value in practice.

In this paper we propose a scheme using image quality metrics and multi-class support vector machine to identify steganographic domains. Firstly, we classify stegnographic domains into four, i.e. spatial domain, DCT domain, DWT domain and ICA domain. Then we analyze total 26 image quality measures summarized by Ismail Avcibas and choose eight sensitive features based on the analysis of variance technique as feature set to distinguish between cover-images and stego-images which are marked in the four domains respectively. The features' scores are computed between the original images and their Gaussian filtered versions. The classifier between cover and stego-images is built using multi-class support vector machine on the selected feature set. The presented method can not only detect the presence of hidden message but also identify the hiding domains. The experiment results show the proposed scheme achieves good classification results and improves its performance with larger embedded message length.

With the explosive growth of mobile data traffic in cellular networks, indoor users are always suffering poor data services. Femtocells are widely accepted as an effective way to solve this problem by providing better coverage. In this way, the user quality of service (QoS) can be significantly enhanced. However, a major obstacle to implement the fashion is lacking market-driven mechanisms to incentivize femtocell owners to trade their access permissions (ACPs). Therefore, designing a robust auction mechanism for ACP trading has attracted lots of attention. A critical challenge of designing such an auction mechanism is to ensure the economic property of truthfulness. Most of the prior works on this issue focus on single-sided scenario, where there is only one seller or one buyer. However, multiple femtocells and multiple macrocell users equipments (MUEs) are always involved in practical systems. In this paper, we study a general market model where multiple femtocells can trade with multiple MUEs and show that designing such a truthful auction mechanism for this scenario is challenging. Therefore, we propose a truthful double auction for access permission (TDAP) allocation. We show analytically that our auction mechanism is economic robust (i.e., satisfying three economic properties including truthfulness, individually rationality, and budget balance) and computationally efficient. Moreover, through extensive simulation experiments, we show TDAP can highly improve auction efficiency outperforming prior auction design.

The paradigm of cloud computing has spontaneously prompted a wide interest in auction-based mechanisms for cloud resource allocation. To eliminate market manipulation, a number of strategy-proof (a.k.a. Truthful) cloud auction mechanisms have been recently proposed by enforcing bidders to bid their true valuations of the cloud resources. However, as discovered in this paper, they would suffer from a new cheating pattern, named false-name bids, where a bidder can gain profit by submitting bids under multiple fictitious names (e.g, Multiple e-mail addresses). Such false-name cheating is easy to make but hard to detect in cloud auctions. To tackle this issue, we propose FAITH, a new False-name-proof Auction for virtual machine instance allocation, that is proven both strategy-proof and false-name proof by our theoretical analysis. When N users compete for M different types of computing instances with multiple units, FAITH achieves a lower time complexity of O(N log N+NM) compared to exiting cloud auction designs. We further extend FAITH to support range-based requests as desired in practice for flexible auction. Through extensive simulation experiments, we show that FAITH highly improves auction efficiency, outperforming the extended mechanisms of conventional false-name-proof auctions in terms of generated revenue and social welfare by up to 220% and 140%, respectively.

Spectrum handoff plays an important role in cognitive radio networks (CRNs). Secondary users (SUs) use spectrum handoff to hold on the idle channel or to free the channel for primary users (PUs). Spectrum handoff scheme greatly affects the transmission quality and the success rate of SUs connection. In this Letter, a reinforcement learning‐based spectrum handoff scheme with the measured packet drop rate (PDR) for multimedia transmissions over CRNs is proposed. In a system model with multiple PUs and SUs, a new state space description method is designed and an observed state includes not only the status whether PUs arrive on each channel but also several other important factors. Also, the measured PDR, instead of the calculated one, is presented to update the mean opinion score, the Q‐table and the handoff policy. Compared with the existing schemes with the calculated PDR from the Quality‐of‐Experience model, the authors' proposed scheme can converge more rapidly in the dynamic radio environment, and reduce the PDR of SUs more significantly.

With the explosion of novel wireless applications,the increasing growth of spectrum requirement has outpaced available spectrum resources.Cognitive Radio Network(CRN) has emerged as a promising solution to address the above dilemma by dynamically sharing spectrum among users.Recently,CRN technology has attracted great research interests as well as efforts.In this paper,we review the state-of-the-art of spectrum allocation techniques in CRNs.We first illustrate the technical background of CRNs,and then analyze the key issues in spectrum allocation algorithms design.Following that we review the design rationale and technical feature of typical allocation models,and further investigate the implementation mechanism of classical algorithms for each model.Finally,we envision the possible issues for future work on spectrum allocation.

ABSTRACT Fruits and vegetable (F&V) account for a large proportion of the market output, and reducing mechanical damage can improve economic benefits. During harvesting and storage of F&V, visible mechanical damage and unseen bruises can result from external loads. However, there are relatively few studies on the micro‐damage of F&V, and the micro‐damage changes are reflected in the macro. Therefore, it is necessary to establish macroscopic, mesoscopic, and microscopic mechanical structure models. In this paper, the factors influencing the mechanics of F&V are analyzed, macroscopic, mesoscopic, and microscopic mechanical analysis methods of F&V are reviewed, multi‐scale mechanical relationships are explored, and future research directions of multi‐scale biomechanics of F&V are prospected. The results show that future research direction should mainly focus on the force analysis of microscopic single cell, as it is the basis of mechanical analysis of F&V.

In wireless networks, a recent trend is to make spectrum access dynamic for efficient utilization of spectrum. In such a scenario, the spectrum is periodically allocated to wireless users using an auction-based market mechanism. A critical property required for designing such a mechanism is truthfulness, which could avoid market manipulation. Such mechanism design typically involves solving NP-hard problems; hence, approximation algorithms are always resorted to in real systems. However, recent results have suggested that it is impossible to implement reasonable approximations without losing robustness to manipulation. In this paper, we solve the problem in a novel perspective by relaxing the constraints of ensuring strong truthfulness. We discuss the concepts of approximate truthfulness and provide approximately truthful mechanisms to improve efficiency (in terms of social welfare and spectrum utilization). We first develop a computationally efficient mechanism that achieves truthful in expectation. This mechanism is based on the assumption that bidders are risk neutral. Following that, we break the assumption by proposing a hard-to-manipulate auction (HMA), which makes it hard to manipulate the auction for profit gains. Our extensive simulation results show that our mechanisms can achieve significant improvement over the state-of-the-art mechanisms.

Abstract In this paper, we employ a well‐known fixed point theorem for cones to study the existence of positive periodic solutions to the n ‐dimensional system x ″ + A ( t ) x = H ( t ) G ( x ). Moreover, the eigenvalue intervals for x ″ + A ( t ) x = λH ( t ) G ( x ) are easily characterized. (© 2008 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

In this article, an ontology cloud model and its corresponding definition and operation is brought forward. The new-knowledge-view and pragmatic which leads to virginity and creativity of knowledge is embodied in this model, and the knowledgepsilas characters including uncertainty, inconsistency, time-varying and sociality, regularity is well expressed in this model too. As a result, Knowledge modeling by ontology-cloud model is a novel idea and foundation of building information, knowledge and intelligent system.

Battle is taking on typical complicated and nonlinear characteristics with the more and more profound application of information technology in military field.The System Dynamics model of the battle in information age based on OODA loop was brought out according to the theory and method of System Dynamics and the process of the battle was simulated.The model reflects better the nature of battle in information age.Result of the simulation will provide the commander with the quantitative theoretical support and policy orientation needed to organize,command and win the war in information condition.

Battle Agent modeling is the key to set up a combat simulation platform based on Agent. Aiming at the implementation problem of the battle Agent model in the combat modeling, the battle Agent modeling method which is based on attribute modeling, capability modeling and behavior mechanism modeling is put forward, through the analysis of tank combat process. On the base, the battle Agent model facing the practicality application is established. In the dynamic evolvement experiment based on the battle agent and the experiment of comparing the battle Agent model and the LAN Chester Equation model, the effectiveness, the feasibility and the operability of the battle Agent are testified, and the battle Agent could satisfy with the requirement of simulating and explaining the combat.