Next Issue
Volume 12, April
Previous Issue
Volume 12, February
 
 

Algorithms, Volume 12, Issue 3 (March 2019) – 16 articles

Cover Story (view full-size image): Large-scale phylogenetic trees—trees depicting the evolutionary history of thousands of species—are often constructed by synthesizing the information in collections of smaller input trees. A basic question here is determining whether the input trees are compatible; i.e., whether the evolutionary relationships they represent are mutually consistent. Compatibility testing relies heavily on data structures for maintaining the connected components of a graph under a series of node and edge deletions. The best-known such data structure, HDT, offers polylogarithmic update time, but incurs considerable overhead because of its complexity. We show that in practice, a much simpler data structure performs better than HDT for compatibility testing, as well as for the closely related incomplete directed perfect phylogeny problem. We present theoretical and empirical justifications for our results. View this paper.
  • Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
  • You may sign up for e-mail alerts to receive table of contents of newly released issues.
  • PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.
Order results
Result details
Section
Select all
Export citation of selected articles as:
11 pages, 1216 KiB  
Article
An Approach to the Dynamics and Control of Uncertain Robot Manipulators
by Xiaohui Yang, Xiaolong Zhang, Shaoping Xu, Yihui Ding, Kun Zhu and Peter Xiaoping Liu
Algorithms 2019, 12(3), 66; https://doi.org/10.3390/a12030066 - 26 Mar 2019
Cited by 8 | Viewed by 4147
Abstract
In this paper, a novel constraint-following control for uncertain robot manipulators that is inspired by analytical dynamics is developed. The motion can be regarded as external constraints of the system. However, it is not easy to obtain explicit equations for dynamic modeling of [...] Read more.
In this paper, a novel constraint-following control for uncertain robot manipulators that is inspired by analytical dynamics is developed. The motion can be regarded as external constraints of the system. However, it is not easy to obtain explicit equations for dynamic modeling of constrained systems. For a multibody system subject to motion constraints, it is a common practice to introduce Lagrange multipliers, but using these to obtain explicit dynamical equations is a very difficult task. In order to obtain such equations more simply, motion constraints are handled here using the Udwadia-Kalaba equation(UKE). Then, considering real-life robot manipulators are usually uncertain(but bounded), by using continuous controllers compensate for the uncertainties. No linearizations/approximations of the robot manipulators systems are made throughout, and the tracking errors are bounds. A redundant manipulator of the SCARA type as the example to illustrates the methodology. Numerical results are demonstrates the simplicity and ease of implementation of the methodology. Full article
Show Figures

Figure 1

17 pages, 6851 KiB  
Article
High-Precision Combined Tidal Forecasting Model
by Jiao Liu, Guoyou Shi and Kaige Zhu
Algorithms 2019, 12(3), 65; https://doi.org/10.3390/a12030065 - 26 Mar 2019
Cited by 12 | Viewed by 5601
Abstract
To improve the overall accuracy of tidal forecasting and ameliorate the low accuracy of single harmonic analysis, this paper proposes a combined tidal forecasting model based on harmonic analysis and autoregressive integrated moving average–support vector regression (ARIMA-SVR). In tidal analysis, the resultant tide [...] Read more.
To improve the overall accuracy of tidal forecasting and ameliorate the low accuracy of single harmonic analysis, this paper proposes a combined tidal forecasting model based on harmonic analysis and autoregressive integrated moving average–support vector regression (ARIMA-SVR). In tidal analysis, the resultant tide can be considered as a superposition of the astronomical tide level and the non-astronomical tidal level, which are affected by the tide-generating force and environmental factors, respectively. The tidal data are de-noised via wavelet analysis, and the astronomical tide level is subsequently calculated via harmonic analysis. The residual sequence generated via harmonic analysis is used as the sample dataset of the non-astronomical tidal level, and the tidal height of the system is calculated by the ARIMA-SVR model. Finally, the tidal values are predicted by linearly summing the calculated results of both systems. The simulation results were validated against the measured tidal data at the tidal station of Bay Waveland Yacht Club, USA. By considering the residual non-astronomical tide level effects (which are ignored in traditional harmonic analysis), the combined model improves the accuracy of tidal prediction. Moreover, the combined model is feasible and efficient. Full article
(This article belongs to the Special Issue Big Data Analytics, Algorithms and Programming)
Show Figures

Figure 1

15 pages, 297 KiB  
Article
A Weighted Voting Ensemble Self-Labeled Algorithm for the Detection of Lung Abnormalities from X-Rays
by Ioannis E. Livieris, Andreas Kanavos, Vassilis Tampakas and Panagiotis Pintelas
Algorithms 2019, 12(3), 64; https://doi.org/10.3390/a12030064 - 16 Mar 2019
Cited by 39 | Viewed by 6367
Abstract
During the last decades, intensive efforts have been devoted to the extraction of useful knowledge from large volumes of medical data employing advanced machine learning and data mining techniques. Advances in digital chest radiography have enabled research and medical centers to accumulate large [...] Read more.
During the last decades, intensive efforts have been devoted to the extraction of useful knowledge from large volumes of medical data employing advanced machine learning and data mining techniques. Advances in digital chest radiography have enabled research and medical centers to accumulate large repositories of classified (labeled) images and mostly of unclassified (unlabeled) images from human experts. Machine learning methods such as semi-supervised learning algorithms have been proposed as a new direction to address the problem of shortage of available labeled data, by exploiting the explicit classification information of labeled data with the information hidden in the unlabeled data. In the present work, we propose a new ensemble semi-supervised learning algorithm for the classification of lung abnormalities from chest X-rays based on a new weighted voting scheme. The proposed algorithm assigns a vector of weights on each component classifier of the ensemble based on its accuracy on each class. Our numerical experiments illustrate the efficiency of the proposed ensemble methodology against other state-of-the-art classification methods. Full article
(This article belongs to the Special Issue Ensemble Algorithms and Their Applications)
Show Figures

Figure 1

17 pages, 1487 KiB  
Article
Synchronization Control Algorithm of Double-Cylinder Forging Hydraulic Press Based on Fuzzy Neural Network
by Xiaodan Xu, Zhifeng Bai and Yuanyuan Shao
Algorithms 2019, 12(3), 63; https://doi.org/10.3390/a12030063 - 14 Mar 2019
Cited by 4 | Viewed by 4603
Abstract
In order to solve the poor control accuracy problem of the traditional synchronous control algorithm for a double-cylinder forging hydraulic press, a synchronous control algorithm for double-cylinder forging hydraulic press based on a fuzzy neural network was proposed. According to the flow equation [...] Read more.
In order to solve the poor control accuracy problem of the traditional synchronous control algorithm for a double-cylinder forging hydraulic press, a synchronous control algorithm for double-cylinder forging hydraulic press based on a fuzzy neural network was proposed. According to the flow equation of valve and hydraulic cylinder, the balance equation and force balance equation of forging hydraulic cylinder are established by using the theory of electro-hydraulic servo systems, and the cylinder-controlled transfer function of forging hydraulic cylinder is deduced. By properly simplifying the transfer function, the mathematical model of synchronous control of double cylinder forging hydraulic press is established. According to the implementation process of traditional fuzzy neural networks, the properties of compensation operation are introduced. The traditional fuzzy neural network is optimized, and the optimized neural network is used to realize the synchronous control of the double cylinder forging hydraulic press. The experimental results show that the amplitude curve of the algorithm is very close to the expected amplitude curve, the error amplitude is only 0.3 mm, and the average control time is about 140 s, which fully shows that the algorithm has high accuracy and a good control effect. Full article
Show Figures

Figure 1

18 pages, 706 KiB  
Article
Multi-View Network Representation Learning Algorithm Research
by Zhonglin Ye, Haixing Zhao, Ke Zhang and Yu Zhu
Algorithms 2019, 12(3), 62; https://doi.org/10.3390/a12030062 - 12 Mar 2019
Cited by 2 | Viewed by 4158
Abstract
Network representation learning is a key research field in network data mining. In this paper, we propose a novel multi-view network representation algorithm (MVNR), which embeds multi-scale relations of network vertices into the low dimensional representation space. In contrast to existing approaches, MVNR [...] Read more.
Network representation learning is a key research field in network data mining. In this paper, we propose a novel multi-view network representation algorithm (MVNR), which embeds multi-scale relations of network vertices into the low dimensional representation space. In contrast to existing approaches, MVNR explicitly encodes higher order information using k-step networks. In addition, we introduce the matrix forest index as a kind of network feature, which can be applied to balance the representation weights of different network views. We also research the relevance amongst MVNR and several excellent research achievements, including DeepWalk, node2vec and GraRep and so forth. We conduct our experiment on several real-world citation datasets and demonstrate that MVNR outperforms some new approaches using neural matrix factorization. Specifically, we demonstrate the efficiency of MVNR on network classification, visualization and link prediction tasks. Full article
Show Figures

Figure 1

15 pages, 1288 KiB  
Article
A Novel Coupling Algorithm Based on Glowworm Swarm Optimization and Bacterial Foraging Algorithm for Solving Multi-Objective Optimization Problems
by Yechuang Wang, Zhihua Cui and Wuchao Li
Algorithms 2019, 12(3), 61; https://doi.org/10.3390/a12030061 - 11 Mar 2019
Cited by 12 | Viewed by 4453
Abstract
In the real word, optimization problems in multi-objective optimization (MOP) and dynamic optimization can be seen everywhere. During the last decade, among various swarm intelligence algorithms for multi-objective optimization problems, glowworm swarm optimization (GSO) and bacterial foraging algorithm (BFO) have attracted increasing attention [...] Read more.
In the real word, optimization problems in multi-objective optimization (MOP) and dynamic optimization can be seen everywhere. During the last decade, among various swarm intelligence algorithms for multi-objective optimization problems, glowworm swarm optimization (GSO) and bacterial foraging algorithm (BFO) have attracted increasing attention from scholars. Although many scholars have proposed improvement strategies for GSO and BFO to keep a good balance between convergence and diversity, there are still many problems to be solved carefully. In this paper, a new coupling algorithm based on GSO and BFO (MGSOBFO) is proposed for solving dynamic multi-objective optimization problems (dMOP). MGSOBFO is proposed to achieve a good balance between exploration and exploitation by dividing into two parts. Part I is in charge of exploitation by GSO and Part II is in charge of exploration by BFO. At the same time, the simulation binary crossover (SBX) and polynomial mutation are introduced into the MGSOBFO to enhance the convergence and diversity ability of the algorithm. In order to show the excellent performance of the algorithm, we experimentally compare MGSOBFO with three algorithms on the benchmark function. The results suggests that such a coupling algorithm has good performance and outperforms other algorithms which deal with dMOP. Full article
Show Figures

Figure 1

21 pages, 941 KiB  
Article
Heterogeneous Distributed Big Data Clustering on Sparse Grids
by David Pfander, Gregor Daiß and Dirk Pflüger
Algorithms 2019, 12(3), 60; https://doi.org/10.3390/a12030060 - 7 Mar 2019
Cited by 4 | Viewed by 5970
Abstract
Clustering is an important task in data mining that has become more challenging due to the ever-increasing size of available datasets. To cope with these big data scenarios, a high-performance clustering approach is required. Sparse grid clustering is a density-based clustering method that [...] Read more.
Clustering is an important task in data mining that has become more challenging due to the ever-increasing size of available datasets. To cope with these big data scenarios, a high-performance clustering approach is required. Sparse grid clustering is a density-based clustering method that uses a sparse grid density estimation as its central building block. The underlying density estimation approach enables the detection of clusters with non-convex shapes and without a predetermined number of clusters. In this work, we introduce a new distributed and performance-portable variant of the sparse grid clustering algorithm that is suited for big data settings. Our computed kernels were implemented in OpenCL to enable portability across a wide range of architectures. For distributed environments, we added a manager–worker scheme that was implemented using MPI. In experiments on two supercomputers, Piz Daint and Hazel Hen, with up to 100 million data points in a ten-dimensional dataset, we show the performance and scalability of our approach. The dataset with 100 million data points was clustered in 1198 s using 128 nodes of Piz Daint. This translates to an overall performance of 352 TFLOPS . On the node-level, we provide results for two GPUs, Nvidia’s Tesla P100 and the AMD FirePro W8100, and one processor-based platform that uses Intel Xeon E5-2680v3 processors. In these experiments, we achieved between 43% and 66% of the peak performance across all computed kernels and devices, demonstrating the performance portability of our approach. Full article
Show Figures

Figure 1

26 pages, 11169 KiB  
Article
Autonomous Population Regulation Using a Multi-Agent System in a Prey–Predator Model That Integrates Cellular Automata and the African Buffalo Optimization Metaheuristic
by Boris Almonacid, Fabián Aspée and Francisco Yimes
Algorithms 2019, 12(3), 59; https://doi.org/10.3390/a12030059 - 6 Mar 2019
Cited by 3 | Viewed by 6004
Abstract
This research focused on the resolution of a dynamic prey–predator spatial model. This model has six life cycles and simulates a theoretical population of prey and predators. Cellular automata represent a set of prey and predators. The cellular automata move in a discrete [...] Read more.
This research focused on the resolution of a dynamic prey–predator spatial model. This model has six life cycles and simulates a theoretical population of prey and predators. Cellular automata represent a set of prey and predators. The cellular automata move in a discrete space in a 2d lattice that has the shape of a torus. African buffaloes represent the predators, and the grasslands represent the prey in the African savanna. Each buffalo moves in the discrete space using the proper motion equation of the African buffalo optimization metaheuristic. Two types of approaches were made with five experiments each. The first approach was the development of a dynamic prey–predator spatial model using the movement of the African buffalo optimization metaheuristic. The second approach added the characteristic of regulating the population of buffaloes using autonomous multi-agents that interact with the model dynamic prey–predator spatial model. According to the obtained results, it was possible to adjust and maintain a balance of prey and predators during a determined period using multi-agents, therefore preventing predators from destroying an entire population of prey in the coexistence space. Full article
Show Figures

Figure 1

14 pages, 305 KiB  
Article
A Selectable Sloppy Heap
by Adrian Dumitrescu
Algorithms 2019, 12(3), 58; https://doi.org/10.3390/a12030058 - 6 Mar 2019
Cited by 5 | Viewed by 4638
Abstract
We study the selection problem, namely that of computing the ith order statistic of n given elements. Here we offer a data structure called selectable sloppy heap that handles a dynamic version in which upon request (i) a new element is inserted [...] Read more.
We study the selection problem, namely that of computing the ith order statistic of n given elements. Here we offer a data structure called selectable sloppy heap that handles a dynamic version in which upon request (i) a new element is inserted or (ii) an element of a prescribed quantile group is deleted from the data structure. Each operation is executed in constant time—and is thus independent of n (the number of elements stored in the data structure)—provided that the number of quantile groups is fixed. This is the first result of this kind accommodating both insertion and deletion in constant time. As such, our data structure outperforms the soft heap data structure of Chazelle (which only offers constant amortized complexity for a fixed error rate 0 < ε 1 / 2 ) in applications such as dynamic percentile maintenance. The design demonstrates how slowing down a certain computation can speed up the data structure. The method described here is likely to have further impact in the field of data structure design in extending asymptotic amortized upper bounds to same formula asymptotic worst-case bounds. Full article
(This article belongs to the Special Issue Efficient Data Structures)
28 pages, 8107 KiB  
Article
Parameter Estimation, Robust Controller Design and Performance Analysis for an Electric Power Steering System
by Van Giao Nguyen, Xuexun Guo, Chengcai Zhang and Xuan Khoa Tran
Algorithms 2019, 12(3), 57; https://doi.org/10.3390/a12030057 - 5 Mar 2019
Cited by 7 | Viewed by 6428
Abstract
This paper presents a parameter estimation, robust controller design and performance analysis for an electric power steering (EPS) system. The parametrical analysis includes the EPS parameters and disturbances, such as the assist motor parameters, sensor-measurement noise, and random road factors, allowing the EPS [...] Read more.
This paper presents a parameter estimation, robust controller design and performance analysis for an electric power steering (EPS) system. The parametrical analysis includes the EPS parameters and disturbances, such as the assist motor parameters, sensor-measurement noise, and random road factors, allowing the EPS stability to be extensively investigated. Based on the loop-shaping technique, the system controller is designed to increase the EPS stability and performance. The loop-shaping procedure is proposed to minimize the influence of system disturbances on the system outputs. The simplified refined instrumental variable (SRIV) algorithm, least squares state variable filter (LSSVF) algorithm and instrumental variable state variable filter (IVSVF) algorithm are applied to reduce the model mismatching between the theoretical EPS models and the real EPS model, as the EPS parameters can be accurately identified based on the experimental EPS data. The performance of the proposed method is thus compared to that of the proportional-integral-derivative (PID) test bench results for the EPS system. The experimental results demonstrated that the proposed loop-shaping controller provides good tracking performance while ensuring the stability of the EPS system. Full article
(This article belongs to the Special Issue Parameter Estimation Algorithms and Its Applications)
Show Figures

Figure 1

23 pages, 4265 KiB  
Article
Matrix Adaptation Evolution Strategy with Multi-Objective Optimization for Multimodal Optimization
by Wei Li
Algorithms 2019, 12(3), 56; https://doi.org/10.3390/a12030056 - 5 Mar 2019
Cited by 4 | Viewed by 4770
Abstract
The standard covariance matrix adaptation evolution strategy (CMA-ES) is highly effective at locating a single global optimum. However, it shows unsatisfactory performance for solving multimodal optimization problems (MMOPs). In this paper, an improved algorithm based on the MA-ES, which is called the matrix [...] Read more.
The standard covariance matrix adaptation evolution strategy (CMA-ES) is highly effective at locating a single global optimum. However, it shows unsatisfactory performance for solving multimodal optimization problems (MMOPs). In this paper, an improved algorithm based on the MA-ES, which is called the matrix adaptation evolution strategy with multi-objective optimization algorithm, is proposed to solve multimodal optimization problems (MA-ESN-MO). Taking advantage of the multi-objective optimization in maintaining population diversity, MA-ESN-MO transforms an MMOP into a bi-objective optimization problem. The archive is employed to save better solutions for improving the convergence of the algorithm. Moreover, the peaks found by the algorithm can be maintained until the end of the run. Multiple subpopulations are used to explore and exploit in parallel to find multiple optimal solutions for the given problem. Experimental results on CEC2013 test problems show that the covariance matrix adaptation with Niching and the multi-objective optimization algorithm (CMA-NMO), CMA Niching with the Mahalanobis Metric and the multi-objective optimization algorithm (CMA-NMM-MO), and matrix adaptation evolution strategy Niching with the multi-objective optimization algorithm (MA-ESN-MO) have overall better performance compared with the covariance matrix adaptation evolution strategy (CMA-ES), matrix adaptation evolution strategy (MA-ES), CMA Niching (CMA-N), CMA-ES Niching with Mahalanobis Metric (CMA-NMM), and MA-ES-Niching (MA-ESN). Full article
Show Figures

Figure 1

13 pages, 5180 KiB  
Article
Depth Optimization Analysis of Articulated Steering Hinge Position Based on Genetic Algorithm
by Bing-wei Cao, Xin-hui Liu, Wei Chen, Yong Zhang and Ai-min Li
Algorithms 2019, 12(3), 55; https://doi.org/10.3390/a12030055 - 5 Mar 2019
Cited by 9 | Viewed by 4616
Abstract
Articulated steering is affected by the position of the articulated points of the steering cylinder. When the two steering cylinders turn, there is a stroke difference and arm of force difference. The existence of the above differences causes the pressure fluctuation of the [...] Read more.
Articulated steering is affected by the position of the articulated points of the steering cylinder. When the two steering cylinders turn, there is a stroke difference and arm of force difference. The existence of the above differences causes the pressure fluctuation of the steering system. Firstly, the mathematical model of the steering mechanism is established through theoretical analysis. Then, the coordinates of the hinge points of the steering cylinder are optimized using genetic algorithm (GA) with the stroke difference function and cylinder pressure function as the objective functions. The curves of the stroke difference and the arm of force difference of the steering cylinder are obtained by mathematical modeling, and the correctness of the GA is verified. According to the optimization results, the wheel loader prototype was reconstructed, and the reconstruction verified by corresponding sensors. The experimental curves show that the steering system has no obvious pressure fluctuation. Finally, the arm of force difference and stroke difference curves were analyzed, and it was concluded that the arm of force difference was the main cause of pressure fluctuation. The objective function was improved, and the arm of force function and cylinder pressure function were taken as the objective functions to continue the optimization by GA. The arm of force difference and stroke difference after optimization were reduced, which provides a constructive reference for the design of articulated steering systems in the future. Full article
Show Figures

Figure 1

21 pages, 3820 KiB  
Article
Parameter Tuning of PI Control for Speed Regulation of a PMSM Using Bio-Inspired Algorithms
by Juan Luis Templos-Santos, Omar Aguilar-Mejia, Edgar Peralta-Sanchez and Raul Sosa-Cortez
Algorithms 2019, 12(3), 54; https://doi.org/10.3390/a12030054 - 4 Mar 2019
Cited by 12 | Viewed by 7149
Abstract
This article focuses on the optimal gain selection for Proportional Integral (PI) controllers comprising a speed control scheme for the Permanent Magnet Synchronous Motor (PMSM). The gains calculation is performed by means of different algorithms inspired by nature, which allows improvement of the [...] Read more.
This article focuses on the optimal gain selection for Proportional Integral (PI) controllers comprising a speed control scheme for the Permanent Magnet Synchronous Motor (PMSM). The gains calculation is performed by means of different algorithms inspired by nature, which allows improvement of the system performance in speed regulation tasks. For the tuning of the control parameters, five optimization algorithms are chosen: Bat Algorithm (BA), Biogeography-Based Optimization (BBO), Cuckoo Search Algorithm (CSA), Flower Pollination Algorithm (FPA) and Sine-Cosine Algorithm (SCA). Finally, for purposes of efficiency assessment, two reference speed profiles are introduced, where an acceptable PMSM performance is attained by using the proposed PI controllers tuned by nature inspired algorithms. Full article
Show Figures

Figure 1

24 pages, 1838 KiB  
Article
Tree Compatibility, Incomplete Directed Perfect Phylogeny, and Dynamic Graph Connectivity: An Experimental Study
by David Fernández-Baca and Lei Liu
Algorithms 2019, 12(3), 53; https://doi.org/10.3390/a12030053 - 28 Feb 2019
Cited by 4 | Viewed by 5375
Abstract
We study two problems in computational phylogenetics. The first is tree compatibility. The input is a collection P of phylogenetic trees over different partially-overlapping sets of species. The goal is to find a single phylogenetic tree that displays all the evolutionary relationships implied [...] Read more.
We study two problems in computational phylogenetics. The first is tree compatibility. The input is a collection P of phylogenetic trees over different partially-overlapping sets of species. The goal is to find a single phylogenetic tree that displays all the evolutionary relationships implied by P . The second problem is incomplete directed perfect phylogeny (IDPP). The input is a data matrix describing a collection of species by a set of characters, where some of the information is missing. The question is whether there exists a way to fill in the missing information so that the resulting matrix can be explained by a phylogenetic tree satisfying certain conditions. We explain the connection between tree compatibility and IDPP and show that a recent tree compatibility algorithm is effectively a generalization of an earlier IDPP algorithm. Both algorithms rely heavily on maintaining the connected components of a graph under a sequence of edge and vertex deletions, for which they use the dynamic connectivity data structure of Holm et al., known as HDT. We present a computational study of algorithms for tree compatibility and IDPP. We show experimentally that substituting HDT by a much simpler data structure—essentially, a single-level version of HDT—improves the performance of both of these algorithm in practice. We give partial empirical and theoretical justifications for this observation. Full article
Show Figures

Figure 1

24 pages, 440 KiB  
Article
Space-Efficient Fully Dynamic DFS in Undirected Graphs
by Kengo Nakamura and Kunihiko Sadakane
Algorithms 2019, 12(3), 52; https://doi.org/10.3390/a12030052 - 27 Feb 2019
Cited by 6 | Viewed by 5075
Abstract
Depth-first search (DFS) is a well-known graph traversal algorithm and can be performed in O ( n + m ) time for a graph with n vertices and m edges. We consider the dynamic DFS problem, that is, to maintain a DFS tree [...] Read more.
Depth-first search (DFS) is a well-known graph traversal algorithm and can be performed in O ( n + m ) time for a graph with n vertices and m edges. We consider the dynamic DFS problem, that is, to maintain a DFS tree of an undirected graph G under the condition that edges and vertices are gradually inserted into or deleted from G. We present an algorithm for this problem, which takes worst-case O ( m n · polylog ( n ) ) time per update and requires only ( 3 m + o ( m ) ) log n bits of space. This algorithm reduces the space usage of dynamic DFS algorithm to only 1.5 times as much space as that of the adjacency list of the graph. We also show applications of our dynamic DFS algorithm to dynamic connectivity, biconnectivity, and 2-edge-connectivity problems under vertex insertions and deletions. Full article
(This article belongs to the Special Issue Efficient Data Structures)
Show Figures

Figure 1

12 pages, 2095 KiB  
Article
Optimized Deep Convolutional Neural Networks for Identification of Macular Diseases from Optical Coherence Tomography Images
by Qingge Ji, Jie Huang, Wenjie He and Yankui Sun
Algorithms 2019, 12(3), 51; https://doi.org/10.3390/a12030051 - 26 Feb 2019
Cited by 82 | Viewed by 6525
Abstract
Finetuning pre-trained deep neural networks (DNN) delicately designed for large-scale natural images may not be suitable for medical images due to the intrinsic difference between the datasets. We propose a strategy to modify DNNs, which improves their performance on retinal optical coherence tomography [...] Read more.
Finetuning pre-trained deep neural networks (DNN) delicately designed for large-scale natural images may not be suitable for medical images due to the intrinsic difference between the datasets. We propose a strategy to modify DNNs, which improves their performance on retinal optical coherence tomography (OCT) images. Deep features of pre-trained DNN are high-level features of natural images. These features harm the training of transfer learning. Our strategy is to remove some deep convolutional layers of the state-of-the-art pre-trained networks: GoogLeNet, ResNet and DenseNet. We try to find the optimized deep neural networks on small-scale and large-scale OCT datasets, respectively, in our experiments. Results show that optimized deep neural networks not only reduce computational burden, but also improve classification accuracy. Full article
Show Figures

Figure 1

Previous Issue
Next Issue
Back to TopTop