Search

One-versus-all (OVA) classifiers learn k individual binary classifiers, each distinguishing the instances of a single class from the instances of all other classes. To classify a new instance, the k classifiers are run, and the one that returns the highest confidence is chosen. Thus, OVA is different from existing data stream classification schemes whose majority use multiclass classifiers, each discriminating among all the classes. This paper advocates some outstanding advantages of OVA for data stream classification. First, there is low error correlation and, hence, high diversity among OVA's component classifiers, which leads to high classification accuracy. Second, OVA is adept at accommodating new class labels that often appear in data streams. However, there also remain many challenges to deploy traditional OVA for classifying data streams. First, traditional OVA does not handle concept change, a key feature of data streams. Second, as every instance is fed to all component classifiers, OVA is known as an inefficient model. Third, OVA's classification accuracy is adversely affected by the imbalanced class distributions in data streams. This paper addresses those key challenges and consequently proposes a new OVA scheme that is adapted for data stream classification. Theoretical analysis and empirical evidence reveal that the adapted OVA can offer faster training, faster updating, and higher classification accuracy than many existing popular data stream classification algorithms. We expect these results to be of interest to researchers and practitioners because they suggest a simple but very elegant and effective alternative to existing classification schemes for data streams. Index Terms--Mining data streams, one-versus-all classifiers.

The purpose of the intrusion detection system (IDS) database is to detect transactions that access data without permission. This paper proposes a novel approach to identifying malicious transactions. The approach concentrates on two aspects of database transactions: (1) dependencies among data items and (2) variations of each individual data item which can be considered as time-series data. The advantages are threefold. First, dependency rules among data items are extended to detect transactions that read or write data without permission. Second, a novel behaviour similarity criterion is introduced to reduce the false positive rate of the detection. Third, time-series anomaly analysis is conducted to pinpoint intrusion transactions that update data items with unexpected pattern. As a result, the proposed approach is able to track normal transactions and detect malicious ones more effectively than existing approaches. [ABSTRACT FROM AUTHOR]

Data intensive peer-to-peer (P2P) networks are becoming increasingly popular in applications like social networking, file sharing networks, etc. Data mining in such P2P environments is the new generation of advanced P2P applications. Unfortunately, most of the existing data mining algorithms do not fit well in such environments since they require data that can be accessed in its entirety. It also is not easy due to the requirements of online transactional data streams. In this paper, we have developed a local algorithm for tracing frequent item sets over a P2P network. The performance of the proposed algorithm is comparatively tested and analyzed through a series of experiments. [ABSTRACT FROM AUTHOR]

Association rule mining is an active data mining research area. Recent years have witnessed many efforts on discovering fuzzy associations. The key strength of fuzzy association rule mining is its completeness. This strength, however, comes with a major drawback to handle large datasets. It often produces a huge number of candidate itemsets. The huge number of candidate itemsets makes it ineffective for a data mining system to analyze them. In the end, it produces a huge number of fuzzy associations. This is particularly true for datasets whose attributes are highly correlated. The huge number of fuzzy associations makes it very difficult for a human user to analyze them. Existing research has shown that most of the discovered rules are actually redundant or insignificant. In this paper, we propose a novel technique to overcome these problems; we are preprocessing the data tuples by focusing on similar behaviour attributes and ontology. Finally, the efficiency and advantages of this algorithm have been proved by experimental results. [ABSTRACT FROM AUTHOR]

Nowadays multi-agent modeling is one of the important problems in problem-domains in which several agents are involved. One of the main reasons for modeling agents is to become capable of predicting their behaviors in special situations. In this paper a modeling algorithm for multi-agent systems is presented which is applicable in domains like RoboCup coach competitions. (a very chaotic and uncertain domain with large number of agents). This algorithm is a combination of a classic data mining algorithm named association rule mining and fuzzy set theory which leads to a soft computing approach. Also some evaluation methods for recognition of models are presented and finally, some experiments for measuring accuracy of models and evaluation methods are outlined. [ABSTRACT FROM AUTHOR]

Most of the problems in the RoboCup soccer domain suffer from the noisy perceptions, noisy actions, and continuous state space. To cope with these problems, using Fuzzy logic can be a proper choice, due to its capabilities of inferring and approximate reasoning under uncertainty. However, designing the entire rule base of a Fuzzy rule base system (FRBS) by an expert is a boring and time consuming task and sometimes the performance of the designed Fuzzy system is far from the optimum, especially in cases that the available knowledge of the system is not enough. In this paper, a rule learning method based on the iterative rule learning (IRL) approach is proposed to generate the entire rule base of an FRBS with the help of genetic algorithms (GAs). The advantage of our proposed method compared to similar approaches in the literature is that our algorithm does not need any training set, which is difficult to collect in many cases; cases like most of the problems existing in the RoboCup soccer domain. As a test case, the goal-shooting problem in the RoboCup 3D soccer simulation league is chosen to be solved using this approach. Simulation tests reveal that with applying the rule learning method proposed in this paper on the goal-shooting problem, not only a rule base with good performance in goal-shooting skill can be obtained, but also the number of rules in the rule base can be decreased by using the general rules in constructing the rule base. [ABSTRACT FROM AUTHOR]

In recent years, classification learning for data streams has become an important and active research topic. A major challenge posed by data streams is that their underlying concepts can change over time, which requires current classifiers to be revised accordingly and timely. To detect concept change, a common method is to observe the online classification accuracy. If accuracy drops below some threshold value, a concept change is deemed to have taken place. An implicit assumption behind this methodology is that any drop in accuracy can be interpreted as a symptom of concept change. Unfortunately however, this assumption is often violated in the real world where data streams carry noise and missing values that can also introduce a significant reduction in classification accuracy. To compound this problem, traditional noise cleansing methods are not applicable to data streams. These methods normally need to scan data multiple times whereas learning in data streams can only afford one-pass scan because of data's high speed and huge volume. To solve these problems, this paper proposes a novel classification algorithm, Class Specific Fuzzy Decision Trees (CSFDT), which utilizes fuzzy logic to classify data streams. The base classifier of CSFDT is a binary fuzzy decision tree. Whenever the problem of concern contains q classes (q > 2), CSFDT learns one binary classifier for each class to distinguish instances of this class from instances of the remaining (q - 1) classes. The CSFDT's advantages are three folds. First, it offers an adaptive structure to effectively and efficiently handle concept change. Second, it is robust to noise. Third, it deals with missing values in an elegant way. As a result, accuracy drop can be safely attributed to concept change. Extensive evaluations are conducted to compare CSFDT with representative existing data stream classification algorithms on a large variety of data. Experimental results suggest that CSFDT provides a significant benefit to data stream classification in real-world scenarios where concept changes, noise and missing values coexist. [ABSTRACT FROM AUTHOR]

With the rise of traffic over wide networks, particularly the internet, and the cloud-based transactions and interactions, database security is important for any organisation. The detection of, and protection from, unauthorised external attacks and insiders abusing privileges is an integral part of database security. To that end, we propose Outlier based Intrusion Detection in Databases for User Behaviour Analysis using Weighted Sequential Pattern Mining (BWSPM), a novel method for the detection of malicious transactions through a sequential flow from outlier detection followed by different behavioural checks at the role-based rule mining component, and finally a user level behavioural check. In the worst case, a transaction has to go through a triple-fold security validation directing the model from generalisation to specification. The Outlier Detection module generates clusters based on the syntactic characteristics of transactions and detects transactions that do not adhere to their closest cluster. Role-level analysis is based upon mining rules that capture dynamic usage of attributes local to every role domain, and the transactions are verified against these rules. Finally, User behaviour profiling models user behaviour based on past transactions, and the incoming transaction is flagged if it diverges from that. Security checks are made at every level to prevent further transaction analysis to reduce false positive rate and achieve a higher degree of optimisation. Encouraging results, with levels of accuracy of around 86.4%, were obtained through our approach after conducting experiments on a dataset generated using the TPC-C (Transaction Processing Performance Council) benchmark. [ABSTRACT FROM AUTHOR]