Biblio

Many standard optimization methods for segmentation and reconstruction compute ML model estimates for appearance or geometry of segments, e.g. Zhu-Yuille [23], Torr [20], Chan-Vese [6], GrabCut [18], Delong et al. [8]. We observe that the standard likelihood term in these formu-lations corresponds to a generalized probabilistic K-means energy. In learning it is well known that this energy has a strong bias to clusters of equal size [11], which we express as a penalty for KL divergence from a uniform distribution of cardinalities. However, this volumetric bias has been mostly ignored in computer vision. We demonstrate signif- icant artifacts in standard segmentation and reconstruction methods due to this bias. Moreover, we propose binary and multi-label optimization techniques that either (a) remove this bias or (b) replace it by a KL divergence term for any given target volume distribution. Our general ideas apply to continuous or discrete energy formulations in segmenta- tion, stereo, and other reconstruction problems.

We propose a dense continuous-time tracking and mapping method for RGB-D cameras. We parametrize the camera trajectory using continuous B-splines and optimize the trajectory through dense, direct image alignment. Our method also directly models rolling shutter in both RGB and depth images within the optimization, which improves tracking and reconstruction quality for low-cost CMOS sensors. Using a continuous trajectory representation has a number of advantages over a discrete-time representation (e.g. camera poses at the frame interval). With splines, less variables need to be optimized than with a discrete representation, since the trajectory can be represented with fewer control points than frames. Splines also naturally include smoothness constraints on derivatives of the trajectory estimate. Finally, the continuous trajectory representation allows to compensate for rolling shutter effects, since a pose estimate is available at any exposure time of an image. Our approach demonstrates superior quality in tracking and reconstruction compared to approaches with discrete-time or global shutter assumptions.

Genes, proteins and other metabolites present in cellular environment exhibit a virtual network that represents the regulatory relationship among its constituents. This network is called Gene Regulatory Network (GRN). Computational reconstruction of GRN reveals the normal metabolic pathway as well as disease motifs. Availability of microarray gene expression data from normal and diseased tissues makes the job easier for computational biologists. Reconstruction of GRN is based on neural modeling. Here we have used discrete and continuous versions of a meta-heuristic algorithm named Firefly algorithm for structure and parameter learning of GRNs respectively. The discrete version for this problem is proposed by us and it has been applied to explore the discrete search space of GRN structure. To evaluate performance of the algorithm, we have used a widely used synthetic GRN data set. The algorithm shows an accuracy rate above 50% in finding GRN. The accuracy level of the performance of Firefly algorithm in structure and parameter optimization of GRN is promising.

Because of poor performance of heuristic algorithms on virtual machine placement problem in cloud environments, a multi-objective constraint optimization model of virtual machine placement is presented, which taking energy consumption and resource wastage as the objective. We solve the model based on the proposed discrete firefly algorithm. It takes firefly's location as the placement result, brightness as the objective value. Its movement strategy makes darker fireflies move to brighter fireflies in solution space. The continuous position after movement is discretized by the proposed discrete strategy. In order to speed up the search for solution, the local search mechanism for the optimal solution is introduced. The experimental results in OpenStack cloud platform show that the proposed algorithm makes less energy consumption and resource wastage compared with other algorithms.

A novel method for computation of modal reflectivity at optical waveguide end-facet is presented. The method is based on the characteristic Green's function (CGF) technique. Using separability assumption of the structure and rational function fitting method (RFFM), a closed-form field expression is derived for optical planar waveguide. The uniform derived expression consists of discrete and continuous spectrum contributions which denotes guided and radiation modes effects respectively. An optimization problem is then defined to calculate the exact reflection coefficients at the end-facet for all extracted poles obtained from rational function fitting step. The proposed CGF-RFFM-optimization offers superior exactness in comparison with the previous reported CGF-complex images (CI) technique due to contribution of all components of field in the optimization problem. The main advantage of the proposed method lies in its simple implementation as well as precision for any refractive index contrast. Excellent numerical agreements with rigorous methods are shown in several examples.

The design of low power chip for IoT applications is very challenge, especially for self-powered wireless sensors. Achieving ultra low power requires both system level optimization and circuit level innovation. This paper presents a continuous-in-time and discrete-in-amplitude (CTDA) system architecture that facilitates adaptive data rate sampling and clockless implementation for a wireless sensor SoC.

We propose a methodology for architecture exploration for Cyber-Physical Systems (CPS) based on an iterative, optimization-based approach, where a discrete architecture selection engine is placed in a loop with a continuous sizing engine. The discrete optimization routine proposes a candidate architecture to the sizing engine. The sizing routine optimizes over the continuous parameters using simulation to evaluate the physical models and to monitor the requirements. To decrease the number of simulations, we show how balance equations and conservation laws can be leveraged to prune the discrete space, thus achieving significant reduction in the overall runtime. We demonstrate the effectiveness of our methodology on an industrial case study, namely an aircraft environmental control system, showing more than one order of magnitude reduction in optimization time.

In practical reflector antenna structures, components of the back-up structure (BUS) are selected form a standard steel library which is normally manufactured. In this case, the design problem of the antenna structure is a discrete optimization problem. In most cases, discrete design is solved by heuristic-based algorithm which will be computing-expensive when the number of deign variable increases. In this paper, a continuous method is used to transfer the discrete optimization problem to a continuous one and gradient-based technique is utilized to solve this problem. The method proposed can achieve a well antenna surface accuracy with all components selected from a standard cross-section list, which is shown by a 9m diameter antenna optimization problem.

In this paper, we focus on the principal-agent problems in continuous time when the participants have ambiguity on the output process in the framework of g-expectation. The first best (or, risk-sharing) type is studied. The necessary condition of the optimal contract is derived by means of the optimal control theory. Finally, we present some examples to clarify our results.

Risk-control optimization has great significance for security of power system. Usually the probabilistic uncertainties of parameters are considered in the research of risk optimization of power system. However, the method of probabilistic uncertainty description will be insufficient in the case of lack of sample data. Thus non-probabilistic uncertainties of parameters should be considered, and will impose a significant influence on the results of optimization. To solve this problem, a robust optimization operation method of power system risk-control is presented in this paper, considering the non-probabilistic uncertainty of parameters based on information gap decision theory (IGDT). In the method, loads are modeled as the non-probabilistic uncertainty parameters, and the model of robust optimization operation of risk-control is presented. By solving the model, the maximum fluctuation of the pre-specified target can be obtained, and the strategy of this situation can be obtained at the same time. The proposed model is applied to the IEEE-30 system of risk-control by simulation. The results can provide the valuable information for operating department to risk management.

In this paper, we investigate the resilient cumulant game control problem for a cyber-physical system. The cyberphysical system is modeled as a linear hybrid stochastic system with full-state feedback. We are interested in 2-player cumulant Nash game for a linear Markovian system with quadratic cost function where the players optimize their system performance by shaping the distribution of their cost function through cost cumulants. The controllers are optimally resilient against control feedback gain variations.We formulate and solve the coupled first and second cumulant Hamilton-Jacobi-Bellman (HJB) equations for the dynamic game. In addition, we derive the optimal players strategy for the second cost cumulant function. The efficiency of our proposed method is demonstrated by solving a numerical example.

Skype has been a typical choice for providing VoIP service nowadays and is well-known for its broad range of features, including voice-calls, instant messaging, file transfer and video conferencing, etc. Considering its wide application, from the viewpoint of ISPs, it is essential to identify Skype flows and thus optimize network performance and forecast future needs. However, in general, a host is likely to run multiple network applications simultaneously, which makes it much harder to classify each and every Skype flow from mixed traffic exactly. Especially, current techniques usually focus on host-level identification and do not have the ability to identify Skype traffic at the flow-level. In this paper, we first reveal the unique sequence signatures of Skype UDP flows and then implement a practical online system named SkyTracer for precise Skype traffic identification. To the best of our knowledge, this is the first time to utilize the strong sequence signatures to carry out early identification of Skype traffic. The experimental results show that SkyTracer can achieve very high accuracy at fine-grained level in identifying Skype traffic.

The longstanding debate on a fundamental science of security has led to advances in systems, software, and network security. However, existing efforts have done little to inform how an environment should react to emerging and ongoing threats and compromises. The authors explore the goals and structures of a new science of cyber-decision-making in the Cyber-Security Collaborative Research Alliance, which seeks to develop a fundamental theory for reasoning under uncertainty the best possible action in a given cyber environment. They also explore the needs and limitations of detection mechanisms; agile systems; and the users, adversaries, and defenders that use and exploit them, and conclude by considering how environmental security can be cast as a continuous optimization problem.
 

Many common cyberdefenses (like firewalls and intrusion-detection systems) are static, giving attackers the freedom to probe them at will. Moving-target defense (MTD) adds dynamism, putting the systems to be defended in motion, potentially at great cost to the defender. An alternative approach is a mobile resilient defense that removes attackers' ability to rely on prior experience without requiring motion in the protected infrastructure. The defensive technology absorbs most of the cost of motion, is resilient to attack, and is unpredictable to attackers. The authors' mobile resilient defense, Ant-Based Cyber Defense (ABCD), is a set of roaming, bio-inspired, digital-ant agents working with stationary agents in a hierarchy headed by a human supervisor. ABCD provides a resilient, extensible, and flexible defense that can scale to large, multi-enterprise infrastructures such as the smart electric grid.

A small battery driven bio-patch, attached to the human body and monitoring various vital signals such as temperature, humidity, heart activity, muscle and brain activity, is an example of a highly resource constrained system, that has the demanding task to assess correctly the state of the monitored subject (healthy, normal, weak, ill, improving, worsening, etc.), and its own capabilities (attached to subject, working sensors, sufficient energy supply, etc.). These systems and many other systems would benefit from a sense of itself and its environment to improve robustness and sensibility of its behavior. Although we can get inspiration from fields like neuroscience, robotics, AI, and control theory, the tight resource and energy constraints imply that we have to understand accurately what technique leads to a particular feature of awareness, how it contributes to improved behavior, and how it can be implemented cost-efficiently in hardware or software. We review the concepts of environment- and self-models, semantic interpretation, semantic attribution, history, goals and expectations, prediction, and self-inspection, how they contribute to awareness and self-awareness, and how they contribute to improved robustness and sensibility of behavior.

This paper presents a novel design of content fingerprints based on maximization of the mutual information across the distortion channel. We use the information bottleneck method to optimize the filters and quantizers that generate these fingerprints. A greedy optimization scheme is used to select filters from a dictionary and allocate fingerprint bits. We test the performance of this method for audio fingerprinting and show substantial improvements over existing learning based fingerprints.

Quasi-steady-state (QSS) large-signal models are often taken for granted in the analysis and design of DC-DC switching converters, particularly for varying operating conditions. In this study, the premise for the QSS is justified quantitatively for the first time. Based on the QSS, the DC-DC switching converter under varying operating conditions is reduced to the linear time varying systems model. Thereafter, the QSS concept is applied to analysis of frequency-domain properties of the DC-DC switching converters by using three-dimensional Bode plots, which is then utilised to the optimisation of the controller parameters for wide variations of input voltage and load resistance. An experimental prototype of an average-current-mode-controlled boost DC-DC converter is built to verify the analysis and design by both frequency-domain and time-domain measurements.

This paper develops an opposition-based learning harmony search algorithm with mutation (OLHS-M) for solving global continuous optimization problems. The proposed method is different from the original harmony search (HS) in three aspects. Firstly, opposition-based learning technique is incorporated to the process of improvisation to enlarge the algorithm search space. Then, a new modified mutation strategy is instead of the original pitch adjustment operation of HS to further improve the search ability of HS. Effective self-adaptive strategy is presented to fine-tune the key control parameters (e.g. harmony memory consideration rate HMCR, and pitch adjustment rate PAR) to balance the local and global search in the evolution of the search process. Numerical results demonstrate that the proposed algorithm performs much better than the existing improved HS variants that reported in recent literature in terms of the solution quality and the stability.

The longstanding debate on a fundamental science of security has led to advances in systems, software, and network security. However, existing efforts have done little to inform how an environment should react to emerging and ongoing threats and compromises. The authors explore the goals and structures of a new science of cyber-decision-making in the Cyber-Security Collaborative Research Alliance, which seeks to develop a fundamental theory for reasoning under uncertainty the best possible action in a given cyber environment. They also explore the needs and limitations of detection mechanisms; agile systems; and the users, adversaries, and defenders that use and exploit them, and conclude by considering how environmental security can be cast as a continuous optimization problem.

This paper presents one-layer projection neural networks based on projection operators for solving constrained variational inequalities and related optimization problems. Sufficient conditions for global convergence of the proposed neural networks are provided based on Lyapunov stability. Compared with the existing neural networks for variational inequalities and optimization, the proposed neural networks have lower model complexities. In addition, some improved criteria for global convergence are given. Compared with our previous work, a design parameter has been added in the projection neural network models, and it results in some improved performance. The simulation results on numerical examples are discussed to demonstrate the effectiveness and characteristics of the proposed neural networks.

Distributed optimization is an emerging research topic. Agents in the network solve the problem by exchanging information which depicts people's consideration on a optimization problem in real lives. In this paper, we introduce two algorithms in continuous-time to solve distributed optimization problems with equality constraints where the cost function is expressed as a sum of functions and where each function is associated to an agent. We firstly construct a continuous dynamic system by utilizing the Lagrangian function and then show that the algorithm is locally convergent and globally stable under certain conditions. Then, we modify the Lagrangian function and re-construct the dynamic system to prove that the new algorithm will be convergent under more relaxed conditions. At last, we present some simulations to prove our theoretical results.

In this paper, we consider distributed algorithm based on a continuous-time multi-agent system to solve constrained optimization problem. The global optimization objective function is taken as the sum of agents' individual objective functions under a group of convex inequality function constraints. Because the local objective functions cannot be explicitly known by all the agents, the problem has to be solved in a distributed manner with the cooperation between agents. Here we propose a continuous-time distributed gradient dynamics based on the KKT condition and Lagrangian multiplier methods to solve the optimization problem. We show that all the agents asymptotically converge to the same optimal solution with the help of a constructed Lyapunov function and a LaSalle invariance principle of hybrid systems.

In this paper, we propose a decomposition based multiobjective evolutionary algorithm that extracts information from an external archive to guide the evolutionary search for continuous optimization problem. The proposed algorithm used a mechanism to identify the promising regions(subproblems) through learning information from the external archive to guide evolutionary search process. In order to demonstrate the performance of the algorithm, we conduct experiments to compare it with other decomposition based approaches. The results validate that our proposed algorithm is very competitive.

An improved harmony search algorithm is presented for solving continuous optimization problems in this paper. In the proposed algorithm, an elimination principle is developed for choosing from the harmony memory, so that the harmonies with better fitness will have more opportunities to be selected in generating new harmonies. Two key control parameters, pitch adjustment rate (PAR) and bandwidth distance (bw), are dynamically adjusted to favor exploration in the early stages and exploitation during the final stages of the search process with the different search spaces of the optimization problems. Numerical results of 12 benchmark problems show that the proposed algorithm performs more effectively than the existing HS variants in finding better solutions.