An Effective Entity Resolution Method.

Iterative Soft Decision Feedback Zig%2DZag Equalizer for 2D Intersymbol Interference Channels.

A Graph%2Dbased Similarity Metric and Validity Indices for Clustering Non%2Dnumeric and Unstructured Data.

Reduced state BCJR algorithms for one%2D and two%2Ddimensional equalization.

Markov Random Field Detection on Two%2DDimensional Intersymbol Interference Channels.

Serially Concatenated IRA Codes

Privacy%2DPreserving Bayesian Network Learning From Heterogeneous Distributed Data.
Abstract — In this paper, we propose a post randomization technique to learn a Bayesian network (BN) from distributed heterogeneous data, in a privacy sensitive fashion. In this case, two or more parties own sensitive data but want to learn a Bayesian network from the combined data. We consider both structure and parameter learning for the BN. The only required information from the data set is a set of sufficient statistics for learning both network structure and parameters. The proposed method estimates the sufficient statistics from the randomized data. The estimated sufficient statistics are then used to learn a BN. For structure learning, we face the familiar extra-link problem since estimation errors tend to break the conditional independence among the variables. We propose modifications of score functions used for BN learning, to solve this problem. We show both theoretically and experimentally that post randomization is an efficient, flexible, and easy-to-use method to learn Bayesian network from privacy sensitive data.
Detection of Markov Random Fields on Two%2DDimensional Intersymbol Interference Channels

Random%2Ddata perturbation techniques and privacy%2Dpreserving data mining.

Privacy%2DSensitive Bayesian Network Parameter Learning.
This paper considers the problem of learning the parameters of a Bayesian Network, assuming the structure of the network is given, from a privacy-sensitive dataset that is distributed between multiple parties. For a binary-valued dataset, we show that the count information required to estimate the conditional probabilities in a Bayesian network can be obtained as a solution to a set of linear equations involving some inner product between the relevant different feature vectors. We consider a random projection-based method that was proposed elsewhere to securely compute the inner product (with a modified implementation of that method). 1.
Collective Mining of Bayesian Networks from Distributed Heterogeneous Data.
We present a collective approach to learning a Bayesian network from distributed heterogenous data. In this approach, we first learn a local Bayesian network at each site using the local data. Then each site identifies the observations that are most likely to be evidence of coupling between local and non-local variables and transmits a subset of these observations to a central site. Another Bayesian network is learnt at the central site using the data transmitted from the local site. The local and central Bayesian networks are combined to obtain a collective Bayesian network, that models the entire data. Experimental results and theoretical justification that demonstrate the feasibility of our approach are presented.
On the Privacy Preserving Properties of Random Data Perturbation Techniques.
Privacy is becoming an increasingly important issue in many data mining applications. This has triggered the development of many privacy-preserving data mining techniques. A large fraction of them use randomized data distortion techniques to mask the data for preserving the privacy of sensitive data. This methodology attempts to hide the sensitive data by randomly modifying the data values often using additive noise. This paper questions the utility of the random value distortion technique in privacy preservation. The paper notes that random objects (particularly random matrices) have “predictable ” structures in the spectral domain and it develops a random matrix-based spectral filtering technique to retrieve original data from the dataset distorted by adding random values. The paper presents the theoretical foundation of this filtering method and extensive experimental results to demonstrate that in many cases random data distortion preserve very little data privacy. 1.
Learning Bayesian Network Structure from Distributed Data.
We propose a collective method to address the problem of learning the structure of a Bayesian network from a distributed heterogeneous data sources. In this case, the dataset is distributed among several sites, with different features at each site. The collective method has four steps: local learning, sample selection, cross learning, and combination of the results. The parents of local nodes can be correctly identified in local learning. The main task of cross learning is to identify the links whose vertices are in different sites (cross links). This is done by transmitting a small subset of samples from each local site to a central site. The combination step involves removing extra links from local Bayesian networks that may be introduced during local learning due to the well known hidden variable problem. The sample selection step selects samples, based on a likelihood criterion, that are possibly evidence of cross links. The overall procedure is called collective learning. Experimental results verify that, for sparsely connected networks, the collective learning method can learn the same structure as that obtained by a centralized learning method (which simply aggregates data from all local sites into a single site). 1
Learning Bayesian Network Structure from Distributed Data.
We propose a collective method to address the problem of learning the structure of a Bayesian network from a distributed heterogeneous data sources. In this case, the dataset is distributed among several sites, with different features at each site. The collective method has four steps: local learning, sample selection, cross learning, and combination of the results. The parents of local nodes can be correctly identified in local learning. The main task of cross learning is to identify the links whose vertices are in different sites (cross links). This is done by transmitting a small subset of samples from each local site to a central site. The combination step involves removing extra links from local Bayesian networks that may be introduced during local learning due to the well known hidden variable problem. The sample selection step selects samples, based on a likelihood criterion, that are possibly evidence of cross links. The overall procedure is called collective learning. Experimental results verify that, for sparsely connected networks, the collective learning method can learn the same structure as that obtained by a centralized learning method (which simply aggregates data from all local sites into a single site). 1
Analysis of privacy preserving random perturbation techniques further explorations.
Privacy is becoming an increasingly important issue in many data mining applications, particularly in the security and defense area. This has triggered the development of many privacy-preserving data mining techniques. A large fraction of them uses randomized data distortion techniques to mask the data for preserving the privacy. They attempt to hide the sensitive data by randomly modifying the data values using additive noise. This paper questions the utility of such randomized data distortion technique for preserving privacy in many cases and urges caution.It notes that random objects (particularly random matrices) have “predictable” structures in the spectral domain and then offers a random matrix-based spectral filtering technique to retrieve original data from the data-set distorted by adding random values. It extends our earlier work questioning the efficacy of random perturbation techniques using additive noise for privacy-preserving data mining in continuous valued domain and presents new results in the discrete domain. It shows that the growing collection of random perturbation-based “privacy-preserving ” data mining techniques may need a careful scrutiny in order to prevent privacy breaches through linear transformations. The paper also presents extensive experimental results in order to support this claim. This is an extension of the paper “Random Data Perturbation
A Random Matrix%2DBased Approach for Dependency Detection from Data Streams.

A New Algorithm for Learning Parameters of a Bayesian Network from Distributed Data.

Dependency Detection in MobiMine and Random Matrices.
This paper describes a novel approach to detect correlation  from data streams in the context of MobiMine --- an experimental mobile  data mining system. It presents a brief description of the MobiMine  and identifies the problem of detecting dependencies among stocks from  incrementally observed financial data streams. This is a non-trivial problem  since the stock-market data is inherently noisy and small incremental  volumes of data makes the estimation process more vulnerable to noise.
Distributed Web Mining Using Bayesian Networks from Multiple Data Streams.

Distributed Clustering Using Collective Principal Component Analysis.
This paper considers distributed clustering of high dimensional heterogeneous data using a distributed  Principal Component Analysis (PCA) technique called the Collective PCA. It presents the Collective  PCA technique that can be used independent of the clustering application. It shows a way to integrate  the Collective PCA with a given o-the-shelf clustering algorithm in order to develop a distributed  clustering technique. It also presents experimental results using dierent test data sets including an  application for web mining.
Asymptotic joint normality of the granulometric moments.

A Morphological Estimator for Clique Potentials of Binary Markov Random Fields.

Collective Principal Component Analysis from Distributed%2C Heterogeneous Data.

A Constant%2Dtime Algorithm for Erosions%2FDilations with Applications to Morphological Texture Feature Computation.
 A computationally efficient algorithm for computing erosions and dilations by one-dimensional grayscale structuring elements with constant slope is proposed. The computational complexity of this algorithm is independent of the size of the support of the structuring function. This is a generalization of the method proposed by Van Herk for the case of erosion and dilation by flat one-dimensional structuring elements. By appropriate combinations of these structuring elements, it is possible to approximate many useful structuring elements. This enables efficient computation of granulometries where the number of operations depends linearly on the number of openings. Theoretical and experimental results comparing the complexity of this algorithm with other standard techniques is presented. Two memory efficient algorithms are then presented. Several implementation issues in computing a granulometry and moments of the associated morphological pattern spectrum are then addressed. An efficient implementation of granulometries for large images on machines with limited memory, by dividing the image into smaller rectangular patches is then discussed. The optimum size of these patches is a function of the specific hardware and has been obtained experimentally for three different hardware platforms. Finally, parallel implementation of the different algorithms on two multiprocessor machines is discussed.
Nonlinear filters in genomic control.

Morphologically Constrained GRFs Applications to Texture Synthesis and Analysis.

Map Estimation of a Class of Morphologically Constrained Gibbs Random Fields.

Segmentation of Mammograms into Distinct Morphological Texture Regions.

Map Estimation of a Class of Morphologically Constrained Gibbs Random Fields.

Morphologically Constrained Gibbs Random Fields.

Discrete morphological size distributions and densities estimation techniques and applications.
Morphological size distributions and densities are frequently used as descriptors of granularity or texture within an image. They have been successfully employed in a number of image processing and analysis tasks, including shape analysis, multiscale shape representation, texture classification, and noise filtering. In most cases however it is not possible to analytically compute these quantities. In this paper, we study the problem of estimating the (discrete) morphological size distribution and density of random images, by means of empirical as well as Monte Carlo estimators. Theoretical and experimental results demonstrate clear superiority of the Monte Carlo estimation approach. Examples illustrate the usefulness of the proposed estimators in traditional image processing and analysis problems. Index Terms --- Granulometries, Markov Random Fields, Mathematical Morphology, Morphological Filtering, Monte Carlo Estimation, Phase Transition, Size Distribution and Density, Texture Classi...
Binary random fields%2C random closed sets%2C and morphological sampling.
We theoretically formulate the problem of processing continuous-space binary random fields by means of mathematical morphology. This may allow us employ mathematical morphology for developing new statistical techniques for the analysis of binary random images. Since morphological transformations of continuous-space binary random fields are not measurable in general, we are naturally forced to employ intermediate steps that require generation of an equivalent random closed set. The relationship between continuous-space binary random fields and random closed sets is thoroughly investigated. As a by-product of this investigation, a number of useful new results, regarding separability of random closed sets, are presented. Our plan, however, suffers from a few technical problems that are prominent in the continuous case. As an alternative, we suggest morphological discretization of binary random fields, random closed sets, as well as morphological operators, thereby effectively implementing...
On the morphological analysis of binary random fields.

Morphological Operators for Image Sequences.

