Mohd. Noor Md. Sap

Anomaly-Based Intrusion Detection using Fuzzy Rough Clustering

It is an important issue for the security of network to detect new intrusion attack and also to increase the detection rates and reduce false positive rates in intrusion detection system (IDS). Anomaly intrusion detection focuses on modeling normal behaviors and identifying significant deviations, which could be novel attacks. The normal and the suspicious behavior in computer networks are hard to predict as the boundaries between them cannot be well defined. We apply the idea of the fuzzy rough c-means (FRCM) to clustering analysis. FRCM integrates the advantage of fuzzy set theory and rough set theory that the improved algorithm to network intrusion detection. The experimental results on dataset KDDCup99 show that our method outperforms the existing unsupervised intrusion detection methods

Rough Set Based Clustering of the Self Organizing Map

The Kohonen Self Organizing Map (SOM) is an excellent tool in exploratory phase of data mining. The SOM is a popular tool that maps a high-dimensional space onto a small number of dimensions by placing similar elements close together, forming clusters. When the number of SOM units is large, to facilitate quantitative analysis of the map and the data, similar units needs to be grouped i.e., clustered. In this paper a two-level clustering based on SOM is proposed, which employs rough set theory to capture the inherent uncertainty involved in cluster analysis. The two-stage procedure (first using SOM to produce the prototypes that are then clustered in the second stage) is found to perform well when compared with crisp clustering of the data and increase the accuracy.

Identifying Zigzag based Perceptually Important Points for indexing financial time series

Financial time series often exhibit high degrees of fluctuation which are considered as noise in time series analysis. To remove noise, several lower bounding the Euclidean distance based dimensionality reduction methods are applied. But, however, these methods do not meet the constraint of financial time series analysis that wants to retain the important points and remove others. Therefore, although a number of methods can retain the important points in the financial time series reduction, but, however, they loss the nature of financial time series which consist of several uptrends, downtrends and sideway trends in different resolutions and in the zigzag directions. In this paper, we propose the Zigzag based Perceptually Important Point Identification method to collect those zigzag movement important points. Further, we propose Zigzag based Multiway Search Tree to index these important points. We evaluate our methods in time series dimensionality reduction. The results show the significant performance comparing to other original method.

Hybrid Kohonen Self Organizing Map for the Uncertainty Involved in Overlapping Clusters Using Simulated Annealing

The Kohonen self organizing map is widely used as a popular tool in the exploratory phase of data mining. The SOM (Self Organizing Maps) maps high dimensional space into a 2-dimensional grid by placing similar elements close together, forming clusters. Recently research experiments presented that to capture the uncertainty involved in cluster analysis, it is not necessary to have crisp boundaries in some clustering operations. In this paper to overcome the uncertainty, an optimized clustering algorithm based on SOM which employs the rough set theory and the Simulated Annealing as a general technique for optimization problems is proposed. The optimized two-level stage SA-Rough SOM (Simulated Annealing – Rough Self Organizing Map) (first using SOM to produce the prototypes that are then clustered in the second stage based on the combination of rough set and simulated annealing) is found to perform well and more accurate compared with the crisp clustering methods (i.e. Incremental SOM) and reduces the errors.

A multiresolution important point retrieval method for financial time series representation

Mining financial time series data without ignoring its characteristics is very important. Financial time series data normally fluctuate unexpectedly which courses very high dimensions. The peak and the dip points of the series may appear frequently over time. These points are known as the most important points which reflect some related events to the market. However, to manipulate financial time series, researchers usually decrease this complexity of time series in their techniques. Consequently, transforming the time series into another easily understanding representation is usually considered as an appropriate approach. In this paper, we propose a multiresolution important point retrieval method for financial time series representation. The idea of the method is based on finding the most important points in multiresolution. These retrieved important points are recorded in each resolution. The collected important points are used to construct the TS-binary search tree. From the TS-binary search tree, the application of time series segmentation is conducted. The experimental results show that the TS-binary search tree representation for financial time series exhibits different performance in different number of cutting points, however, in the empirical results, the number of cutting points which are larger than 12 points show the better results.

An intelligent system based on kernel methods for crop yield prediction

This paper presents work on developing a software system for predicting crop yield from climate and plantation data. At the core of this system is a method for unsupervised partitioning of data for finding spatio-temporal patterns in climate data using kernel methods which offer strength to deal with complex data. For this purpose, a robust weighted kernel k-means algorithm incorporating spatial constraints is presented. The algorithm can effectively handle noise, outliers and auto-correlation in the spatial data, for effective and efficient data analysis, and thus can be used for predicting oil-palm yield by analyzing various factors affecting the yield.

An Optimized Hybrid Kohonen Neural Network for Ambiguity Detection in Cluster Analysis Using Simulated Annealing

One of the popular tools in the exploratory phase of Data mining and Pattern Recognition is the Kohonen Self Organizing Map (SOM). The SOM maps the input space into a 2-dimensional grid and forms clusters. Recently experiments represented that to catch the ambiguity involved in cluster analysis, it is not necessary to have crisp boundaries in some clustering operations. In this paper to overcome the ambiguity involved in cluster analysis, a combination of Rough set Theory and Simulated Annealing is proposed that has been applied on the output grid of SOM. Experiments show that the proposed two-stage algorithm, first using SOM to produce the prototypes then applying rough set and SA in the second stage in order to assign the overlapped data to true clusters they belong to, outperforms the proposed crisp clustering algorithms (i.e. I-SOM) and reduces the errors.

Using rough set based multi-checkpointing for fault-tolerance scheduling in economic grids

Fault tolerant Grid scheduling is of vital importance in the Grid computing world. Task replication and checkpointing is two popular methods to achieve a fault tolerant scheduling. Replication method is not an applicable way in economic-based grid computing due to use a large number of resources. The cost of spent time must be paid by consumer for all participant nodes. In this paper, we proposed a fault-tolerant scheduling technique based on Multi-Checkpointing by using rough set theory for economic-based grid with respect to minimum cost, high efficiency, and minimum latency. In our proposed approach, we assume that if one of the provider nodes is failed, there is not enough time to start a task on a new node from beginning again. The experimental results show a promising method with less computation cost price and better fault-tolerance in acceptable completion time.

A new rough set based approach for optimized detection of available resources in Grid computing

Since the Grid is a dynamic environment, the prediction and detection of available resources in near future is important for resource scheduling. Economic-based grid management has been viewed as a feasible approach to carry out fair, efficient and reliable scheduling. One key issue in Economic-based grid strategy is to inform about available resources. In this paper, we present a novel predictable method to specify available resource in economic-based grid. This method use a rough set analysis by scheduler to divide resources in groups and then grant a priority to each group based on cost price and efficiency of nodes. The result show that our proposed method has an acceptable performance and it try to use cheaper and suitable resources for each job to decrease cost price of computation.