scholarly journals Accelerating Protein Structure Prediction using Particle Swarm Optimization on GPU

2015 ◽  
Author(s):  
Hamed Khakzad ◽  
Yasaman Karami ◽  
Seyed Shahriar ARAB
Keyword(s):  
Particle Swarm Optimization ◽  
Structure Prediction ◽  
Tertiary Structure ◽  
Fitness Function ◽  
Particle Swarm ◽  
Search Space ◽  
Processing Unit ◽  
Multiple Sequence ◽  
Swarm Optimization ◽  
Acceleration Rate

Protein tertiary structure prediction (PSP) is one of the most challenging problems in bioinformatics. Different methods have been introduced to solve this problem so far, but PSP is computationally intensive and belongs to the NP-hard class. One of the best solutions to accelerate PSP is the use of a massively parallel processing architecture, such graphical processing unit (GPU), which is used to parallelize computational algorithms. In this paper, we have proposed a parallel architecture to accelerate PSP. A bio-inspired method, particle swarm optimization (PSO) has been used as the optimization method to solve PSP. We have also performed a comprehensive study on implementing different topologies of PSO on GPU to consider the acceleration rate. Our solution belongs to ab-initio category which is based on the dihedral angles and calculates the energy-levels to predict the tertiary structure. Indeed, we have studied the search space of a protein to find the best pair of angles that gives the minimum free energy. A profile-level knowledge-based force field based on PSI-BLAST multiple sequence alignment has been applied as a fitness function to calculate the energy values. Different topologies and variations of PSO are considered here and the experimental results show that the speedup gain using GPU is about 34 times faster than CPU implementation of the algorithm with an acceptable precision. The energy values of predicted structures confirm the robustness of the algorithm.

scholarly journals A New Representation in PSO for Discretization-Based Feature Selection

2021 ◽  
Author(s):  
B Tran ◽  
Bing Xue ◽  
Mengjie Zhang
Keyword(s):  
Feature Selection ◽  
Particle Swarm Optimization ◽  
Fitness Function ◽  
Particle Swarm ◽  
Search Space ◽  
Classification Performance ◽  
High Dimensional ◽  
Swarm Optimization ◽  
Learning Capacity ◽  
Feature Interactions

In machine learning, discretization and feature selection (FS) are important techniques for preprocessing data to improve the performance of an algorithm on high-dimensional data. Since many FS methods require discrete data, a common practice is to apply discretization before FS. In addition, for the sake of efficiency, features are usually discretized individually (or univariate). This scheme works based on the assumption that each feature independently influences the task, which may not hold in cases where feature interactions exist. Therefore, univariate discretization may degrade the performance of the FS stage since information showing feature interactions may be lost during the discretization process. Initial results of our previous proposed method [evolve particle swarm optimization (EPSO)] showed that combining discretization and FS in a single stage using bare-bones particle swarm optimization (BBPSO) can lead to a better performance than applying them in two separate stages. In this paper, we propose a new method called potential particle swarm optimization (PPSO) which employs a new representation that can reduce the search space of the problem and a new fitness function to better evaluate candidate solutions to guide the search. The results on ten high-dimensional datasets show that PPSO select less than 5% of the number of features for all datasets. Compared with the two-stage approach which uses BBPSO for FS on the discretized data, PPSO achieves significantly higher accuracy on seven datasets. In addition, PPSO obtains better (or similar) classification performance than EPSO on eight datasets with a smaller number of selected features on six datasets. Furthermore, PPSO also outperforms the three compared (traditional) methods and performs similar to one method on most datasets in terms of both generalization ability and learning capacity. © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

scholarly journals A New Representation in PSO for Discretization-Based Feature Selection

2021 ◽  
Author(s):  
B Tran ◽  
Bing Xue ◽  
Mengjie Zhang
Keyword(s):  
Feature Selection ◽  
Particle Swarm Optimization ◽  
Fitness Function ◽  
Particle Swarm ◽  
Search Space ◽  
Classification Performance ◽  
High Dimensional ◽  
Swarm Optimization ◽  
Learning Capacity ◽  
Feature Interactions

In machine learning, discretization and feature selection (FS) are important techniques for preprocessing data to improve the performance of an algorithm on high-dimensional data. Since many FS methods require discrete data, a common practice is to apply discretization before FS. In addition, for the sake of efficiency, features are usually discretized individually (or univariate). This scheme works based on the assumption that each feature independently influences the task, which may not hold in cases where feature interactions exist. Therefore, univariate discretization may degrade the performance of the FS stage since information showing feature interactions may be lost during the discretization process. Initial results of our previous proposed method [evolve particle swarm optimization (EPSO)] showed that combining discretization and FS in a single stage using bare-bones particle swarm optimization (BBPSO) can lead to a better performance than applying them in two separate stages. In this paper, we propose a new method called potential particle swarm optimization (PPSO) which employs a new representation that can reduce the search space of the problem and a new fitness function to better evaluate candidate solutions to guide the search. The results on ten high-dimensional datasets show that PPSO select less than 5% of the number of features for all datasets. Compared with the two-stage approach which uses BBPSO for FS on the discretized data, PPSO achieves significantly higher accuracy on seven datasets. In addition, PPSO obtains better (or similar) classification performance than EPSO on eight datasets with a smaller number of selected features on six datasets. Furthermore, PPSO also outperforms the three compared (traditional) methods and performs similar to one method on most datasets in terms of both generalization ability and learning capacity. © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

scholarly journals Embedding parts in shape grammars using a parallel particle swarm optimization method on graphics processing units

2018 ◽  
Vol 32 (3) ◽  
pp. 256-268 ◽  
Author(s):  
Hacer Yalim Keles
Keyword(s):  
Particle Swarm Optimization ◽  
Graphics Processing Units ◽  
Particle Swarm ◽  
Search Space ◽  
Optimization Method ◽  
Processing Unit ◽  
Shape Grammars ◽  
Swarm Optimization ◽  
Parallel Particle Swarm Optimization ◽  
Graphics Processing

AbstractEmbedding emergent parts in shape grammars is computationally challenging. The first challenge is the representation of shapes, which needs to enable reinterpretation of parts regardless of the creation history of the shapes. The second challenge is the relevant part searching algorithm that provides an extensive exploration of the design space–time efficiently. In this work, we propose a novel method to solve both problems; we treat shapes as they are and use a parallel particle swarm optimization-based algorithm to compute emergent parts. The execution time of the proposed method is improved substantially by dividing the search space into small parts and carrying out searches in each part concurrently using a graphics processing unit. The experiments show that the proposed implementation detects emergent parts accurately and time efficiently.

Localization of WSN using Distributed Particle Swarm Optimization algorithm with precise references

2016 ◽  
Vol 7 ◽  
pp. 1
Author(s):  
Ravichander Janapati ◽  
Ch. Balaswamy ◽  
K. Soundararajan
Keyword(s):  
Particle Swarm Optimization ◽  
Search Algorithm ◽  
Particle Swarm ◽  
Search Space ◽  
Research Area ◽  
Population Based ◽  
Wireless Sensor ◽  
Cooperative Localization ◽  
Localization Algorithm ◽  
Swarm Optimization

Localization is the key research area in wireless sensor networks. Finding the exact position of the node is known as localization. Different algorithms have been proposed. Here we consider a cooperative localization algorithm with censoring schemes using Crammer Rao bound (CRB). This censoring scheme  can improve the positioning accuracy and reduces computation complexity, traffic and latency. Particle swarm optimization (PSO) is a population based search algorithm based on the swarm intelligence like social behavior of birds, bees or a school of fishes. To improve the algorithm efficiency and localization precision, this paper presents an objective function based on the normal distribution of ranging error and a method of obtaining the search space of particles. In this paper  Distributed localization of wireless sensor networksis proposed using PSO with best censoring technique using CRB. Proposed method shows better results in terms of position accuracy, latency and complexity.  

scholarly journals Estimation of Fuzzy Parameters in the Linear Muskingum Model with the Aid of Particle Swarm Optimization

2021 ◽  
Vol 13 (13) ◽  
pp. 7152
Author(s):  
Mike Spiliotis ◽  
Alvaro Sordo-Ward ◽  
Luis Garrote
Keyword(s):  
Particle Swarm Optimization ◽  
Fitness Function ◽  
Particle Swarm ◽  
Flood Routing ◽  
Performance Criteria ◽  
Extension Principle ◽  
Swarm Optimization ◽  
Muskingum Model ◽  
Muskingum Method ◽  
Fuzzy Parameters

The Muskingum method is one of the widely used methods for lumped flood routing in natural rivers. Calibration of its parameters remains an active challenge for the researchers. The task has been mostly addressed by using crisp numbers, but fuzzy seems a reasonable alternative to account for parameter uncertainty. In this work, a fuzzy Muskingum model is proposed where the assessment of the outflow as a fuzzy quantity is based on the crisp linear Muskingum method but with fuzzy parameters as inputs. This calculation can be achieved based on the extension principle of the fuzzy sets and logic. The critical point is the calibration of the proposed fuzzy extension of the Muskingum method. Due to complexity of the model, the particle swarm optimization (PSO) method is used to enable the use of a simulation process for each possible solution that composes the swarm. A weighted sum of several performance criteria is used as the fitness function of the PSO. The function accounts for the inclusive constraints (the property that the data must be included within the produced fuzzy band) and for the magnitude of the fuzzy band, since large uncertainty may render the model non-functional. Four case studies from the references are used to benchmark the proposed method, including smooth, double, and non-smooth data and a complex, real case study that shows the advantages of the approach. The use of fuzzy parameters is closer to the uncertain nature of the problem. The new methodology increases the reliability of the prediction. Furthermore, the produced fuzzy band can include, to a significant degree, the observed data and the output of the existent crisp methodologies even if they include more complex assumptions.

Method of Optimization for Target Localization Model Parameters Based on LSSVR

2011 ◽  
Vol 268-270 ◽  
pp. 934-939
Author(s):  
Xue Wen He ◽  
Gui Xiong Liu ◽  
Hai Bing Zhu ◽  
Xiao Ping Zhang
Keyword(s):  
Particle Swarm Optimization ◽  
Fitness Function ◽  
Particle Swarm ◽  
Parameters Optimization ◽  
Target Localization ◽  
Likelihood Method ◽  
Support Vector ◽  
Swarm Optimization ◽  
Localization Accuracy ◽  
Localization Method

Aiming at improving localization accuracy in Wireless Sensor Networks (WSN) based on Least Square Support Vector Regression (LSSVR), making LSSVR localization method more practicable, the mechanism of effects of the kernel function for target localization based on LSSVR is discussed based on the mathematical solution process of LSSVR localization method. A novel method of modeling parameters optimization for LSSVR model using particle swarm optimization is proposed. Construction method of fitness function for modeling parameters optimization is researched. In addition, the characteristics of particle swarm parameters optimization are analyzed. The computational complexity of parameters optimization is taken into consideration comprehensively. Experiments of target localization based on CC2430 show that localization accuracy using LSSVR method with modeling parameters optimization increased by 23%~36% in compare with the maximum likelihood method(MLE) and the localization error is close to the minimum with different LSSVR modeling parameters. Experimental results show that adapting a reasonable fitness function for modeling parameters optimization using particle swarm optimization could enhance the anti-noise ability significantly and improve the LSSVR localization performance.

Sign in / Sign up

Export Citation Format

Share Document