The paper introduces the subject and scope of planned and partly implemented works during the preparation of the dissertation. The topic of establishing parameters of heuristic methods through machine learning, undert...
详细信息
In this paper, we use a well-known Deep Learning technique called Long Short Term Memory (LSTM) recurrent neural networks to find sessions that are prone to code failure in applications that rely on telemetry data for...
详细信息
ISBN:
(纸本)9781728100593
In this paper, we use a well-known Deep Learning technique called Long Short Term Memory (LSTM) recurrent neural networks to find sessions that are prone to code failure in applications that rely on telemetry data for system health monitoring. We also use LSTM networks to extract telemetry patterns that lead to a specific code failure. For code failure prediction, we treat the telemetry events, sequence of telemetry events and the outcome of each sequence as words, sentence and sentiment in the context of sentiment analysis, respectively. Our proposed method is able to process a large set of data and can automatically handle edge cases in code failure prediction. We take advantage of Bayesian optimization technique to find the optimal hyper parameters as well as the type of LSTM cells that leads to the best prediction performance. We then introduce the Contributors and Blockers concepts. In this paper, contributors are the set of events that casue a code failure, while blockers are the set of events that each of them individually prevents a code failure from happening, even in presence of one or multiple contributor(s). Once the proposed LSTM model is trained, we use a greedy approach to find the contributors and blockers. To develop and test our proposed method, we use synthetic (simulated) data in the first step. The synthetic data is generated using a number of rules for code failures, as well as a number of rules for preventing a code failure from happening. The trained LSTM model shows over 99% accuracy for detecting code failures in the synthetic data. The results from the proposed method outperform the classical learning models such as Decision Tree and Random Forest. Using the proposed greedy method, we are able to find the contributors and blockers in the synthetic data in more than 90% of the cases, with a performance better than sequential rule and pattern mining algorithms. In the next step, we train and test our proposed LSTM method on real data that we
An inverted pendulum system has potential applications in different domains that motivate researchers for new innovative development. An inverted pendulum system is an underactuated, nonlinear, inherently unstable and...
详细信息
ISBN:
(数字)9781728162249
ISBN:
(纸本)9781728162256
An inverted pendulum system has potential applications in different domains that motivate researchers for new innovative development. An inverted pendulum system is an underactuated, nonlinear, inherently unstable and a multivariable system. The system is modelled mainly through either Euler-Lagrange or Newtonian dynamic formulation. This paper aims to examine the control of a double inverted pendulum (DIP) using pole placement and linear quadratic regulator (LQR) control. To find the optimal parameters of the LQR control law, Genetic Algorithm (GA) and Particle Swarm optimization (PSO) are used to tune and determine the proper control parameters. Simulations are conducted using MATLAB/Simulink under different circumstances and the performance of each control technique is analyzed and compared in terms of the system rise time, settling time, peak amplitude, and steady state error.
The Internet of Things (IoT) has found wide applications in human endeavors, and has recently drawn massive attention in the field of cognitive science by connecting the Brain-Computer Interface (BCI) to the cloud. Th...
详细信息
ISBN:
(数字)9781728148922
ISBN:
(纸本)9781728148939
The Internet of Things (IoT) has found wide applications in human endeavors, and has recently drawn massive attention in the field of cognitive science by connecting the Brain-Computer Interface (BCI) to the cloud. The integration of EEG-based BCI with internet is rapidly developing due to its non-invasiveness, and the benefits are inexhaustible such as control of assistive robots, telemedicine and telehealth for secured monitoring. However, some challenges exist which would prevent the application of BCIs, including the accurately predicting movement intention from EEG signals, computational complexity while preprocessing EEG tasks, and the problem of overfitting. Several algorithms have been used to address these challenges, but they are still limited to be applied efficiently on the BCI-based IoT. In this paper, in order to tackle the above mentioned issues, we proposed an extended Particle Swarm optimization (PSO)-based neural network (NN) which is able to provide a seamless interaction between the BCI and IoT devices. In the experiments, we designed a BCI system that captured relevant EEG information by means of the PSO, and then projected the features into a neural network system for training. The experimental results demonstrated the feasibility of the proposed PSO-based NN technique in classifying the motor imagery (MI) tasks with an accuracy of 98.9%. Some possible improvement in future work was also suggested.
In this paper, we consider the 3D beamforming with Multi-Active Multi-Passive (MAMP) antenna arrays. For the optimization of the hybrid array's active elements' (AEs) weights and passive elements' (PEs) lo...
详细信息
ISBN:
(数字)9781728154787
ISBN:
(纸本)9781728154794
In this paper, we consider the 3D beamforming with Multi-Active Multi-Passive (MAMP) antenna arrays. For the optimization of the hybrid array's active elements' (AEs) weights and passive elements' (PEs) loads, we propose a novel algorithm, i.e., the alternating optimization - stochastic 3D beamforming algorithm (AO-S3DBA). The scheme is built on a generalized cost function and alternates between the optimization of the loads and weights. Despite the dramatic increase in data points, we managed to decrease the algorithm's complexity by employing Adam, a popular accelerator for stochastic optimization, for the update of the optimization variables. We present simulation results, where the proposed MAMP array can successfully emulate the beam of a uniform rectangular array (URA) with the use of 50% less AEs and with beam steering capabilities towards various azimuth and elevation directions. We believe that this newly proposed type of hybrid transceiver offers a good trade-off between cost and performance, which can be particularly useful in many industrial and Internet-of-Things (IoT) applications, where the deployment of a large number of transceivers / sensors directly affects the application's total cost.
Successful project activities in the IT industry are determined by the extent of difficulty in the team formation and implementation of projects themselves. IT projects provide for fulfillment of a number of tasks tha...
详细信息
Successful project activities in the IT industry are determined by the extent of difficulty in the team formation and implementation of projects themselves. IT projects provide for fulfillment of a number of tasks that are interrelated. In formation of such a project, account should be taken of certain factors necessary for its successful implementation, determination of the technology for fulfilling intermediate tasks: consecutive or parallel, and for setting the priorities. This approach requires detailed calculation and scientifically grounded decisions. The authors have proposed an original approach to solving discrete optimization problems related to fundamental calculation difficulties in the process of an IT project formation. The known methods of exact or approximate solution of such problems are studied with account taken of their belonging to so-called P- and NP-class problems (the polynomial and the exponential solution algorithms). The modern combinatory and heuristic methods for solving practical discrete optimization problems require development of algorithms that allow obtaining approximate solution with guaranteed estimate of deviation from the optimum. Simplification algorithms provide an efficient method of searching for an optimization problem solution. Should a multidimensional process be projected onto the two-dimensional surface, this will enable graphical visualization of sets of the problem solutions. This research provides a way for simplifying the combinatory solution of a discrete optimization problem. It is based on decomposition of the system that represents the system constraining a multidimensional output problem to the two-dimensional coordinate plane. Such method allows obtaining a simple system of graphical solutions of a complicated linear discrete optimization problem. From the practical point of view, the proposed method allows reducing the calculation complicacy of optimization problems belonging to this class when the IT projec
The feasibility of provisioning physical layer security via reconfigurable intelligent surfaces (RISs) is investigated. The key idea is to constructively combine the signals received through the direct and reflected c...
详细信息
ISBN:
(数字)9781728154787
ISBN:
(纸本)9781728154794
The feasibility of provisioning physical layer security via reconfigurable intelligent surfaces (RISs) is investigated. The key idea is to constructively combine the signals received through the direct and reflected channels at a legitimate user, while degrading the signal quality at a passive eavesdropper by destructive combining of these two signals enabled via a joint effect of intelligent phase control at a large number of passive reflective elements of the RIS and a corresponding precoder design of the base-station. To this end, we advocate adoption of linear precoders, which are designed based on cascaded channels, with sole optimization of phase-shifts at the RIS to maximize the achievable secrecy rate in the presence of a passive eavesdropper. The underlying optimization problem is solved by developing an alternating algorithm to iteratively update the phase-shifts of the RIS. Specifically, the optimal phase-shift design problem can be modeled as a semidefinite program by using relaxation techniques. Thereby, the optimal phase-shift values can be closely approximated by using Gaussian randomization techniques. Our numerical results show that the proposed technique serves as a practical-viable low-complexity alternative to joint optimization of non-linear precoder and RIS phase-shifts.
In order to provide users with reliable and qualified power, it becomes an indispensable task to enhance the forecasting capability of the short-term power load. However, the existing approaches of short-term electric...
详细信息
ISBN:
(纸本)9781728140698
In order to provide users with reliable and qualified power, it becomes an indispensable task to enhance the forecasting capability of the short-term power load. However, the existing approaches of short-term electric load forecasting are not proper enough. A short-term electric load forecasting method based on grey neural network based on snap-drift cuckoo search optimization algorithm(SDCS-GNN) is proposed in this paper. Parameters of gray neural network (GNN) are selected randomly which is similar to the initial spatial position of birds' eggs in the parasitic nest of cuckoo. The SDCS is utilized to search the better weight and threshold of the conventional gray neural network (GNN), which improves the stability and accuracy of the prediction model. To validate the superior performance of the proposed method, several well-known evolutionary algorithms such as particle swarm optimization (PSO), grey wolf optimization(GWO), moth-fire suppression optimization(MFO) and cuckoo search optimization (CS) are employed to constitute the contrast experiment of the prediction of short-term power load. The mean squared error predicted by the SDCS-GNN model is the smallest, which compared with GNN, PSO-GNN, GWO-GNN, MFO-GNN, and CS-GNN is 0.36, 1.79, 15.23, 4.53, 2.93, respectively. The Average prediction accuracy of SDCS-GNN model is better than other models which is 7.1592, 1.427, 15.1516, 11.5438, 10.5202, respectively. The simulation results show that the SDCS-GNN model has better approximation ability and higher prediction accuracy than the conventional GNN and other evolutionary algorithms in the short-term electric load forecasting. The experiments above indicates that the prediction method is effective and feasible.
The proceedings contain 32 papers. The special focus in this conference is on algorithms and Computations. The topics include: Flat-Foldability for 1 × n Maps with Square/Diagonal Grid Patterns;(k,p)-Planarity: A...
ISBN:
(纸本)9783030105631
The proceedings contain 32 papers. The special focus in this conference is on algorithms and Computations. The topics include: Flat-Foldability for 1 × n Maps with Square/Diagonal Grid Patterns;(k,p)-Planarity: A Relaxation of Hybrid Planarity;drawing Clustered Graphs on Disk Arrangements;computing the Metric Dimension by Decomposing Graphs into Extended Biconnected Components: (Extended Abstract);on the Algorithmic Complexity of Double Vertex-Edge Domination in Graphs;The Upper Bound on the Eulerian Recurrent Lengths of Complete Graphs Obtained by an IP Solver;a Fast Algorithm for Unbounded Monotone Integer Linear Systems with Two Variables per Inequality via Graph Decomposition;multilevel Planarity;weighted Upper Edge Cover: Complexity and Approximability;parameterized Computational Geometry via Decomposition Theorems;linear Pseudo-Polynomial Factor Algorithm for Automaton Constrained Tree Knapsack Problem;matching Sets of Line Segments;efficient Algorithm for Box Folding;analyzing the Quantum Annealing Approach for Solving Linear Least Squares Problems;greedy Consensus Tree and Maximum Greedy Consensus Tree Problems;a Two Query Adaptive Bitprobe Scheme Storing Five Elements;applications of V-Order: Suffix Arrays, the Burrows-Wheeler Transform & the FM-index;towards Work-Efficient Parallel Parameterized algorithms;arbitrary Pattern Formation on Infinite Grid by Asynchronous Oblivious Robots;packing 2D Disks into a 3D Container;r-Gatherings on a Star;covering and Packing of Rectilinear Subdivision;minimum Membership Covering and Hitting;capacitated Discrete Unit Disk Cover;topological Stability of Kinetic k-centers;a Linear Time Algorithm for the r-Gathering Problem on the Line (Extended Abstract);maximum-Width Empty Square and Rectangular Annulus;hard and Easy Instances of L-Tromino Tilings;the Prefix Fréchet Similarity.
Precipitation is one of the most important elements in meteorological data. However, due to the limitation of resource conditions, the number of meteorological stations is limited, and interpolation is required to obt...
详细信息
ISBN:
(纸本)9781728140698
Precipitation is one of the most important elements in meteorological data. However, due to the limitation of resource conditions, the number of meteorological stations is limited, and interpolation is required to obtain the precipitation data in the observation area and other locations. Kriging interpolation whose core is to obtain the best variogram model is widely used in the prediction of regional ***, it is difficult to find perfect estimating model, and numerous approachs are utilized to handle this problem. In order to gain better parameters and model, an improved spatio-temporal Kriging interpolation method is proposed in this paper. The chaotic ant-lion algorithm (CALO) is employed to seek suitable parameters of the variogram both in the space domain and the time domain. This evolutionary algorithm whose performance has been validated in the literatures is not vulnerable to search the global solution. The experiment is conducted in terms of the fitting effect and interpolation effect, error analysis to demonstrate the superior performance of the proposed method, compared to other fitting methods such as Least square method. Several optimizationalgorithms are used to constitute the contrast *** experimental results show that the proposed method prevails among other approachs as far as the precision, calculation cost and effectiveness.
暂无评论