A New Optimal Binary Tree SVM Multi-Class Classification Algorithm

2013 ◽  
Vol 373-375 ◽  
pp. 1085-1088 ◽  
Author(s):  
Yu Ping Qin ◽  
Peng Da Qin ◽  
Yi Wang ◽  
Shu Xian Lun
Keyword(s):  
Binary Tree ◽  
Classification Algorithm ◽  
Experimental Results ◽  
Leaf Node ◽  
Distribution Area ◽  
Root Node ◽  
Ellipsoid Volume ◽  
Multi Class Classification

A improved binary tree SVM multi-class classification algorithm is proposed. Firstly, constructing the minimum hyper ellipsoid for each class sample in the feather space, and then generating optimal binary tree according to the hyper ellipsoid volume, training sub-classifier for every non-leaf node in the binary tree at the same time. For the sample to be classified, the sub-classifiers are used from the root node until one leaf node, and the corresponding class of the leaf node is the class of the sample. The experiments are done on the Statlog database, and the experimental results show that the algorithm improves classification precision and classification speed, especially in the situation that the number of class are more and their distribution area are equal approximately, the algorithm can greatly improve the classification precision and classification speed.

A New SVM Multi-Class Classification Algorithm Based on Sample Scale and Distribution Area

2013 ◽  
Vol 712-715 ◽  
pp. 2529-2533
Author(s):  
Yu Ping Qin ◽  
Peng Da Qin ◽  
Shu Xian Lun ◽  
Yi Wang
Keyword(s):  
Binary Tree ◽  
Classification Performance ◽  
Classification Algorithm ◽  
Experimental Results ◽  
Leaf Node ◽  
Distribution Area ◽  
Root Node ◽  
Multi Class Classification

A new SVM multi-class classification algorithm is proposed. Firstly, the optimal binary tree is constructed by the scale and the distribution area of every class sample, and then the sub-classifiers are trained for every non-leaf node in the binary tree. For the sample to be classified, the classification is done from the root node until someone leaf node, and the corresponding class of the leaf node is the class of the sample. The experimental results show that the algorithm improves the classification precision and classification speed, especially in the situation that the sample scale is less but its distribution area is bigger, the algorithm can improve greatly the classification performance.

Color Quantization Based on Hierarchical Frequency Sensitive Competitive Learning

2010 ◽  
Vol 14 (4) ◽  
pp. 375-381
Author(s):  
Jun Zhang ◽  
◽  
Jinglu Hu
Keyword(s):  
Binary Tree ◽  
Competitive Learning ◽  
Tree Structure ◽  
Experimental Results ◽  
Color Quantization ◽  
Adaptive Procedure ◽  
Root Node ◽  
Binary Tree Structure

In this paper, we propose a Hierarchical Frequency Sensitive Competitive Learning (HFSCL) method to achieve Color Quantization (CQ). In HFSCL, the appropriate number of quantized colors and the palette can be obtained by an adaptive procedure following a binary tree structure with nodes and layers. Starting from the root node that contains all colors in an image until all nodes are examined by split conditions, a binary tree will be generated. In each node of the tree, a Frequency Sensitive Competitive Learning (FSCL) network is used to achieve two-way division. To avoid over-split, merging condition is defined to merge the clusters that are close enough to each other at each layer. Experimental results show that the proposed HFSCL has desired ability for CQ.

scholarly journals A Generic Algorithm to Determine Maximum Bottleneck Node Weight-based Data Gathering Trees for Wireless Sensor Networks

2015 ◽  
Vol 7 (3) ◽  
pp. 18 ◽  
Author(s):  
Natarajan Meghanathan
Keyword(s):  
Wireless Sensor Networks ◽  
Sensor Networks ◽  
Data Gathering ◽  
Leaf Node ◽  
Wireless Sensor ◽  
Generic Algorithm ◽  
Intermediate Node ◽  
Root Node ◽  
Bottleneck Node ◽  
Node Weight

We propose a generic algorithm to determine maximum bottleneck node weight-based data gathering (MaxBNW-DG) trees for wireless sensor networks (WSNs) and compare the performance of the MaxBNW-DG trees with those of maximum and minimum link weight-based data gathering trees (MaxLW-DG and MinLW-DG trees). Assuming each node in a WSN graph has a weight, the bottleneck weight for the path from a node u to the root node of the DG tree is the minimum of the node weights on the path (inclusive of the weights of the end nodes). The MaxBNW-DG tree algorithm determines a DG tree such that each node has a path of the largest bottleneck weight to the root node. We observe the MaxBNW-DG trees to incur lower height, larger percentage of nodes as leaf nodes and a larger weight per intermediate node compared to the leaf node; the tradeoff being a larger a network-wide data aggregation delay due to larger number of child nodes per intermediate node. The MaxBNW-DG algorithm could be used to determine DG trees with larger trust score, larger energy (and other such criterion for node weight) per intermediate node compared to the leaf node. 

Automatic Ship Routing with High Reliability and Efficiency between Two Arbitrary Points at Sea

2018 ◽  
Vol 72 (2) ◽  
pp. 430-446
Author(s):  
Shuaidong Jia ◽  
Zeyuan Dai ◽  
Lihua Zhang
Keyword(s):  
Binary Tree ◽  
High Reliability ◽  
Spatial Database ◽  
Experimental Results ◽  
Ship Routing ◽  
Shortest Distance ◽  
Tree Index ◽  
Tree Method

Due to the limitations of the existing methods (for example, the route binary tree method) that can only automatically generate routes based on a single chart, a method for automatically generating the shortest distance route based on an obstacle spatial database is proposed. Using this proposed method, the route between two arbitrary points at sea can be automatically generated. First, the differences in accuracy and updating time of charts are quantitatively analysed. Next, the mechanism for updating obstacles is designed, an obstacle spatial database is constructed, and the obstacle data extracted from multiple charts are fused. Finally, considering the effect of efficiency on the amount of obstacle data, a route window and an improved R-tree index are designed for quickly extracting and querying the obstacle database. The experimental results demonstrate that compared with existing methods, the proposed method can generate the shortest distance between two arbitrary points at sea and eliminates the limitation of the area of the chart. In addition, with data from multiple charts, the route generated by the proposed method is more reliable than that of the existing methods, and it is more efficient.

Geometric Figure Recognition by Using Multiple Algorithms Combination

2011 ◽  
Vol 474-476 ◽  
pp. 417-421
Author(s):  
Jia Wei Xu ◽  
Seop Hyeong Park ◽  
Xian Yun Fei
Keyword(s):  
User Interface ◽  
Graphical User Interface ◽  
Classification Algorithm ◽  
Experimental Results ◽  
Geometric Figure ◽  
Geometric Shapes ◽  
Recognition Algorithms ◽  
Multiple Algorithms

This thesis is mainly focused on the geometric figure recognition. We provided many diverse shapes such as rectangle, ellipse, square, circle, parallelogram and other shapes. Based on different kinds of geometric shapes, a number of commonly used figure classification algorithm were designed to recognize all kinds of figures and a geometric figure painted on palette arbitrarily. The experimental results indicate that figure recognition algorithms can be well performed in an integrated graphical user interface.

scholarly journals A Novel Mathematical Formula for Retrieval Algorithm

2014 ◽  
Vol 2014 ◽  
pp. 1-5
Author(s):  
Yuping Qin ◽  
Hamid Reza Karimi ◽  
Aihua Zhang ◽  
Qiangkui Leng
Keyword(s):  
Binary Tree ◽  
Experimental Results ◽  
Retrieval Algorithm ◽  
Mathematical Formula ◽  
Structure Code

A method is proposed to retrieve mathematical formula in LaTeX documents. Firstly, we represent the retrieved mathematical formula by binary tree according to its LaTeX description, normalize the structure of the binary tree, and obtain the structure code and then search the mathematical formula table that is named by the structure code and the formula elements of the first two levels of the binary tree in the mathematical formula database. If the table exists, then we search the normalizing variable name preorder traversing sequence of the binary tree in the table and display the document information that contain the mathematical formula. The experimental results show that the algorithm realizes the retrieval of mathematical formula in LaTeX documents and has higher retrieval precision and faster retrieval speed.

Classification Methods in the Detection of New Suspicious Emails

2008 ◽  
Vol 07 (03) ◽  
pp. 209-217 ◽  
Author(s):  
S. Appavu Alias Balamurugan ◽  
G. Athiappan ◽  
M. Muthu Pandian ◽  
R. Rajaram
Keyword(s):  
Neural Network ◽  
Data Mining ◽  
Decision Tree ◽  
Binary Tree ◽  
Experimental Results ◽  
Classification Methods ◽  
Detection Rates ◽  
Naive Bayesian ◽  
The Past ◽  
Naïve Bayesian

Email has become one of the fastest and most economical forms of communication. However, the increase of email users has resulted in the dramatic increase of suspicious emails during the past few years. This paper proposes to apply classification data mining for the task of suspicious email detection based on deception theory. In this paper, email data was classified using four different classifiers (Neural Network, SVM, Naïve Bayesian and Decision Tree). The experiment was performed using weka on the basis of different data size by which the suspicious emails are detected from the email corpus. Experimental results show that simple ID3 classifier which make a binary tree, will give a promising detection rates.

A Balanced Multiway Search Tree for Multi-Dimension Searching

2010 ◽  
Vol 44-47 ◽  
pp. 3574-3578
Author(s):  
Ai Guo Li ◽  
Chi Zhang ◽  
Jiu Long Zhang ◽  
Zhen Hai Zhang
Keyword(s):  
Search Tree ◽  
Experimental Results ◽  
Extra Cost ◽  
Leaf Node ◽  
Index Structure ◽  
Data Sets ◽  
Comprehensive Performance ◽  
Data Files ◽  
Tree Index ◽  
Tree Building

A new multi-dimensional index structure called RSR-tree is proposed, which based on RS-tree. In RSR-tree, index records of a leaf node are split to ensure the sequence ordering of index records in a leaf node, which reduces the addressing cost of I/O operations effectively when reading data files. The entries of a non-leaf node are split to decreases the overlap between the brother nodes, which reduces effectively the time of reading data from data files. Experimental results on different data sets show that compared to RS-tree, RSR-tree has better comprehensive performance, in regard to tree building and querying. The querying performance is increased and extra cost is not produced.

Sign in / Sign up

Export Citation Format

Share Document