Speeding up AdaBoost Classifier with Random Projection

The development of techniques for scaling up classifiers so that they can be applied to problems with large datasets of training examples is one of the objectives of data mining. Recently, AdaBoost has become popular among machine learning community thanks to its promising results across a variety of applications. However, training AdaBoost on large datasets is a major problem, especially when the dimensionality of the data is very high. This paper discusses the effect of high dimensionality on the training process of AdaBoost. Two preprocessing options to reduce dimensionality, namely the principal component analysis and random projection are briefly examined. Random projection subject to a probabilistic length preserving transformation is explored further as a computationally light preprocessing step. The experimental results obtained demonstrate the effectiveness of the proposed training process for handling high dimensional large datasets.

Usefulness of DARPA dataset for intrusion detection system evaluation

The MIT Lincoln Laboratory IDS evaluation methodology is a practical solution in terms of evaluating the performance of Intrusion Detection Systems, which has contributed tremendously to the research progress in that field. The DARPA IDS evaluation dataset has been criticized and considered by many as a very outdated dataset, unable to accommodate the latest trend in attacks. Then naturally the question arises as to whether the detection systems have improved beyond detecting these old level of attacks. If not, is it worth thinking of this dataset as obsolete? The paper presented here tries to provide supporting facts for the use of the DARPA IDS evaluation dataset. The two commonly used signature-based IDSs, Snort and Cisco IDS, and two anomaly detectors, the PHAD and the ALAD, are made use of for this evaluation purpose and the results support the usefulness of DARPA dataset for IDS evaluation.

Cluster based core vector machine

Core Vector Machine(CVM) is suitable for efficient large-scale pattern classification. In this paper, a method for improving the performance of CVM with Gaussian kernel function irrespective of the orderings of patterns belonging to different classes within the data set is proposed. This method employs a selective sampling based training of CVM using a novel kernel based scalable hierarchical clustering algorithm. Empirical studies made on synthetic and real world data sets show that the proposed strategy performs well on large data sets.

Minimum enclosing spheres formulations for support vector ordinal regression

We present two new support vector approaches for ordinal regression. These approaches find the concentric spheres with minimum volume that contain most of the training samples. Both approaches guarantee that the radii of the spheres are properly ordered at the optimal solution. The size of the optimization problem is linear in the number of training samples. The popular SMO algorithm is adapted to solve the resulting optimization problem. Numerical experiments on some real-world data sets verify the usefulness of our approaches for data mining.

Reducing the babel in plant volatile communication: using the forest to see the trees

While plants of a single species emit a diversity of volatile organic compounds (VOCs) to attract or repel interacting organisms, these specific messages may be lost in the midst of the hundreds of VOCs produced by sympatric plants of different species, many of which may have no signal content. Receivers must be able to reduce the babel or noise in these VOCs in order to correctly identify the message. For chemical ecologists faced with vast amounts of data on volatile signatures of plants in different ecological contexts, it is imperative to employ accurate methods of classifying messages, so that suitable bioassays may then be designed to understand message content. We demonstrate the utility of `Random Forests' (RF), a machine-learning algorithm, for the task of classifying volatile signatures and choosing the minimum set of volatiles for accurate discrimination, using datam from sympatric Ficus species as a case study. We demonstrate the advantages of RF over conventional classification methods such as principal component analysis (PCA), as well as data-mining algorithms such as support vector machines (SVM), diagonal linear discriminant analysis (DLDA) and k-nearest neighbour (KNN) analysis. We show why a tree-building method such as RF, which is increasingly being used by the bioinformatics, food technology and medical community, is particularly advantageous for the study of plant communication using volatiles, dealing, as it must, with abundant noise.

On Finding the Natural Number of Topics with Latent Dirichlet Allocation: Some Observations

It is important to identify the ``correct'' number of topics in mechanisms like Latent Dirichlet Allocation(LDA) as they determine the quality of features that are presented as features for classifiers like SVM. In this work we propose a measure to identify the correct number of topics and offer empirical evidence in its favor in terms of classification accuracy and the number of topics that are naturally present in the corpus. We show the merit of the measure by applying it on real-world as well as synthetic data sets(both text and images). In proposing this measure, we view LDA as a matrix factorization mechanism, wherein a given corpus C is split into two matrix factors M-1 and M-2 as given by C-d*w = M1(d*t) x Q(t*w).Where d is the number of documents present in the corpus anti w is the size of the vocabulary. The quality of the split depends on ``t'', the right number of topics chosen. The measure is computed in terms of symmetric KL-Divergence of salient distributions that are derived from these matrix factors. We observe that the divergence values are higher for non-optimal number of topics - this is shown by a `dip' at the right value for `t'.

Noise and outlier removal from jet engine health signals using weighted FIR median hybrid filters

The removal of noise and outliers from measurement signals is a major problem in jet engine health monitoring. Topical measurement signals found in most jet engines include low rotor speed, high rotor speed. fuel flow and exhaust gas temperature. Deviations in these measurements from a baseline 'good' engine are often called measurement deltas and the health signals used for fault detection, isolation, trending and data mining. Linear filters such as the FIR moving average filter and IIR exponential average filter are used in the industry to remove noise and outliers from the jet engine measurement deltas. However, the use of linear filters can lead to loss of critical features in the signal that can contain information about maintenance and repair events that could be used by fault isolation algorithms to determine engine condition or by data mining algorithms to learn valuable patterns in the data, Non-linear filters such as the median and weighted median hybrid filters offer the opportunity to remove noise and gross outliers from signals while preserving features. In this study. a comparison of traditional linear filters popular in the jet engine industry is made with the median filter and the subfilter weighted FIR median hybrid (SWFMH) filter. Results using simulated data with implanted faults shows that the SWFMH filter results in a noise reduction of over 60 per cent compared to only 20 per cent for FIR filters and 30 per cent for IIR filters. Preprocessing jet engine health signals using the SWFMH filter would greatly improve the accuracy of diagnostic systems. (C) 2002 Published by Elsevier Science Ltd.

Dynamic subspace clustering for very large high-dimensional databases

Emerging high-dimensional data mining applications needs to find interesting clusters embeded in arbitrarily aligned subspaces of lower dimensionality. It is difficult to cluster high-dimensional data objects, when they are sparse and skewed. Updations are quite common in dynamic databases and they are usually processed in batch mode. In very large dynamic databases, it is necessary to perform incremental cluster analysis only to the updations. We present a incremental clustering algorithm for subspace clustering in very high dimensions, which handles both insertion and deletions of datapoints to the backend databases.

A fast algorithm for finding frequent episodes in event streams

Frequent episode discovery is a popular framework for mining data available as a long sequence of events. An episode is essentially a short ordered sequence of event types and the frequency of an episode is some suitable measure of how often the episode occurs in the data sequence. Recently,we proposed a new frequency measure for episodes based on the notion of non-overlapped occurrences of episodes in the event sequence, and showed that, such a definition, in addition to yielding computationally efficient algorithms, has some important theoretical properties in connecting frequent episode discovery with HMM learning. This paper presents some new algorithms for frequent episode discovery under this non-overlapped occurrences-based frequency definition. The algorithms presented here are better (by a factor of N, where N denotes the size of episodes being discovered) in terms of both time and space complexities when compared to existing methods for frequent episode discovery. We show through some simulation experiments, that our algorithms are very efficient. The new algorithms presented here have arguably the least possible orders of spaceand time complexities for the task of frequent episode discovery.

Clustering Based Large Margin Classification: A Scalable Approach using SOCP Formulation

This paper presents a novel Second Order Cone Programming (SOCP) formulation for large scale binary classification tasks. Assuming that the class conditional densities are mixture distributions, where each component of the mixture has a spherical covariance, the second order statistics of the components can be estimated efficiently using clustering algorithms like BIRCH. For each cluster, the second order moments are used to derive a second order cone constraint via a Chebyshev-Cantelli inequality. This constraint ensures that any data point in the cluster is classified correctly with a high probability. This leads to a large margin SOCP formulation whose size depends on the number of clusters rather than the number of training data points. Hence, the proposed formulation scales well for large datasets when compared to the state-of-the-art classifiers, Support Vector Machines (SVMs). Experiments on real world and synthetic datasets show that the proposed algorithm outperforms SVM solvers in terms of training time and achieves similar accuracies.

A unified view of the Apriori-based algorithms for frequent episode discovery

Frequent episode discovery framework is a popular framework in temporal data mining with many applications. Over the years, many different notions of frequencies of episodes have been proposed along with different algorithms for episode discovery. In this paper, we present a unified view of all the apriori-based discoverymethods for serial episodes under these different notions of frequencies. Specifically, we present a unified view of the various frequency counting algorithms. We propose a generic counting algorithm such that all current algorithms are special cases of it. This unified view allows one to gain insights into different frequencies, and we present quantitative relationships among different frequencies.Our unified view also helps in obtaining correctness proofs for various counting algorithms as we show here. It also aids in understanding and obtaining the anti-monotonicity properties satisfied by the various frequencies, the properties exploited by the candidate generation step of any apriori-based method. We also point out how our unified view of counting helps to consider generalization of the algorithm to count episodes with general partial orders.

Discovering injective episodes with general partial orders

Frequent episode discovery is a popular framework for temporal pattern discovery in event streams. An episode is a partially ordered set of nodes with each node associated with an event type. Currently algorithms exist for episode discovery only when the associated partial order is total order (serial episode) or trivial (parallel episode). In this paper, we propose efficient algorithms for discovering frequent episodes with unrestricted partial orders when the associated event-types are unique. These algorithms can be easily specialized to discover only serial or parallel episodes. Also, the algorithms are flexible enough to be specialized for mining in the space of certain interesting subclasses of partial orders. We point out that frequency alone is not a sufficient measure of interestingness in the context of partial order mining. We propose a new interestingness measure for episodes with unrestricted partial orders which, when used along with frequency, results in an efficient scheme of data mining. Simulations are presented to demonstrate the effectiveness of our algorithms.

System and Method for Temporal Datamining

A system for temporal data mining includes a computer readable medium having an application configured to receive at an input module a temporal data series and a threshold frequency. The system is further configured to identify, using a candidate identification and tracking module, one or more occurrences in the temporal data series of a candidate episode and increment a count for each identified occurrence. The system is also configured to produce at an output module an output for those episodes whose count of occurrences results in a frequency exceeding the threshold frequency.

Generalized frequent episodes in Event Sequences

Discovering patterns in temporal data is an important task in Data Mining. A successful method for this was proposed by Mannila et al. [1] in 1997. In their framework, mining for temporal patterns in a database of sequences of events is done by discovering the so called frequent episodes. These episodes characterize interesting collections of events occurring relatively close to each other in some partial order. However, in this framework(and in many others for finding patterns in event sequences), the ordering of events in an event sequence is the only allowed temporal information. But there are many applications where the events are not instantaneous; they have time durations. Interesting episodesthat we want to discover may need to contain information regarding event durations etc. In this paper we extend Mannila et al.’s framework to tackle such issues. In our generalized formulation, episodes are defined so that much more temporal information about events can be incorporated into the structure of an episode. This significantly enhances the expressive capability of the rules that can be discovered in the frequent episode framework. We also present algorithms for discovering such generalized frequent episodes.

A unified view of the apriori-based algorithms for frequent episode discovery

Frequent episode discovery framework is a popular framework in temporal data mining with many applications. Over the years, many different notions of frequencies of episodes have been proposed along with different algorithms for episode discovery. In this paper, we present a unified view of all the apriori-based discovery methods for serial episodes under these different notions of frequencies. Specifically, we present a unified view of the various frequency counting algorithms. We propose a generic counting algorithm such that all current algorithms are special cases of it. This unified view allows one to gain insights into different frequencies, and we present quantitative relationships among different frequencies. Our unified view also helps in obtaining correctness proofs for various counting algorithms as we show here. It also aids in understanding and obtaining the anti-monotonicity properties satisfied by the various frequencies, the properties exploited by the candidate generation step of any apriori-based method. We also point out how our unified view of counting helps to consider generalization of the algorithm to count episodes with general partial orders.