Dimitrios Papadopoulos Dept. of Computer Science and Engineering, UC Riverside, Riverside, CA 92521 edu!ucr!cs!dimitris http://www.cs.ucr.edu/~dimitris/ Education * University of California Riverside, Riverside, CA. PhD in Computer Science (completed on Jan 3, 2005) Thesis title: Clustering and Indexing Methods for High Dimensional Data and Moving Objects * University of California Riverside, Riverside, CA. M.S. in Computer Science, Aug. 2003 Project title: Clustering Gene Expression Data in SQL Using Locally Adaptive Metrics * University of Ioannina, Greece B.S. in Computer Science, Sep. 1998 Work * Research Assistant Database Lab, CS Dept. Experience Univ. California Riverside Spring 2001 Winter 2005 Advisor: Prof. Dimitrios Gunopulos * Teaching Assistant CS Dept. Univ. California Riverside F1999, W2000, S2000, S2001, W2004 Classes: CS8 Introduction to Computing, CS10/CS12 Introduction to Computer Science (C++ / Visual Studio .NET), CS130 Computer Graphics (OpenGL), CS141 Algorithms * Software Engineer MEDLAB, E.U. Project TEMeTeN Univ. of Ioannina, Greece. Sep. 1998 Sep. 1999. Led the group that implemented the back-end application server and configured the database servers of a distributed computerbased patient record system. Publications * Carlotta Domeniconi, Dimitrios Gunopulos, Sheng Ma, Bojun Yan, Muna Al-Razgan, Dimitris Papadopoulos: "Locally adaptive metrics for clustering high dimensional data", Data Min. Knowl. Discov. 14(1): 63-97 (2007) * Maria Halkidi, Vana Kalogeraki, Dimitrios Gunopulos, Dimitris Papadopoulos, Demetris Zeinalipour-Yazti, Michalis Vlachos: "Efficient Online State Tracking Using Sensor Networks", 7th International Conference on Mobile Data Management (MDM'06), Nara, Japan, May 2006 * Sharmila Subramaniam, Themis Palpanas, Dimitris Papadopoulos, Vana Kalogeraki, Dimitrios Gunopulos: "Online Outlier Detection in Sensor Data Using Non-Parametric Models", In Proc. VLDB 2006: 187-198 * Maria Halkidi, Dimitris Papadopoulos, Vana Kalogeraki, Dimitrios Gunopulos: "Resilient and Energy Efficient Tracking in Sensor Networks", International Journal of Wireless and Mobile Computing (accepted) * George Kollios, Dimitris Papadopoulos, Dimitrios Gunopulos, Vassilis J. Tsotras: "Indexing Mobile Objects Using Dual Transformations", The VLDB Journal Vol. 14(2): 238-256 (Apr. 2005) (Online First, Sep. 2004) * Carlotta Domeniconi, Dimitris Papadopoulos, Dimitrios Gunopulos, Sheng Ma: "Subspace Clustering of High Dimensional Data", SIAM International Conference on Data Mining (SDM), Apr. 2004 * Themistoklis Palpanas, Dimitris Papadopoulos, Vana Kalogeraki, Dimitrios Gunop- ulos: "Distributed Deviation Detection in Sensor Networks", SIGMOD Record, Vol 32, No. 4, Dec. 2003 * Dimitris Papadopoulos, Carlotta Domeniconi, Dimitrios Gunopulos, Sheng Ma: "Clustering Gene Expression Data in SQL Using Locally Adaptive Metrics", 8th ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery (DMKD), Jun. 2003 * Dimitris Papadopoulos, George Kollios, Dimitrios Gunopulos, Vassilis J. Tsotras: "Indexing Mobile Objects on the Plane", 5th DEXA International Workshop on Mobility in Databases and Distributed Systems (MDDS), Sep. 2002 * Dimitris Papadopoulos, George Kollios, Dimitrios Gunopulos, Vassilis J. Tsotras: "Indexing Mobile Objects Using Duality Transforms", IEEE Data Engineering Bul- letin 25(2): 18-24 (Jun. 2002) Research My research interests fall into the area of managing and processing multi-dimensional Interests data. In particular, I have been involved in projects related to multidimensional indexing techniques, clustering algorithms, data mining in static databases, as well as in sensor networks and streaming environments. An outline of current and future paths of interests follows: * The task of indexing moving objects in order to answer various spatio-temporal queries inherently involves handling and processing data with many attributes. A problem like "Give the objects' ID's which will be in a particular area A during a specific time interval t in the future" appears in real-life applications such as pre- dicting future congestion areas in a highway system or allocating more bandwidth for areas where a high volume of mobile phone usage is imminent. * Clustering is another data mining task, during which the high dimensionality of the data makes it hard to be applied. One promising approach is to focus on a different, possibly overlapping, subset of the objects' features in order to form each cluster, i.e. perform subspace clustering. In datasets where each object is described by hundreds or thousands of attributes (e.g. document corpora like the REUTERS dataset), it is extremely unlikely that the objects are correlated along each attribute. In these cases, subspace clustering algorithms try to identify clus- ters, which are formed on a subset of dimensions. These techniques can be ap- plied to a diverse set of domains. For instance, biologists seek to identify genes which are co-expressed under certain conditions or stimuli. Subspace clustering algorithms can tackle this problem efficiently. In the telecommunication industry, call-detail records offer a wealth of customer behavior information. Identifying sets of customers that share behavior patterns is another useful application of subspace clustering algorithms. * Data mining tasks are usually harder to be carried out when the data is processed in a streaming fashion. The unique nature of datastream processing ("You got to see the data only once!"), along with computational constraints which are often inherent in such settings (e.g. small memory of sensors), contribute to the dif- ficulty of devising such data mining frameworks. In the setting of sensornets, I am interested into developing data mining and in-network processing techniques, which function in a distributed and online fashion, and can be efficiently deployed on memory, CPU, and power constrained devices. Projects * Implementation of clustering algorithms in SQL [DMKD'03] and C/C++ [SDM'04, DMKD'07] * C/C++ implementation of moving object indexing techniques in external memory [VLDBJ'05, MDDS'02] * Implementation of outlier detection techniques for sensor networks; evaluation and simulation in Java [VLDB'06] * Implementation of distributed algorithms for target tracking in sensornets; evalu- ation and simulation in Java [IJWMC'06, MDM'06] * Shared Storage API library, multi-threaded server and client application: The server provided storage services, while supporting locking, caching, user authenti- cation, encryption and Unicode string support. * Implemented various compiler optimization techniques, using C++ and STL. * Measurements of multicast tree characteristics: Focused on the characteristics of multicast trees in IP Multicast, i.e. out-degree of nodes and distances between re- ceivers. Built graph visualization tool using the GFC toolkit by IBM Alphaworks. * TPC-H benchmark on DB2 installations under Linux and Windows NT. * C++ implementation of the Apriori algorithm for finding frequent itemsets. * Design and implementation of the back-end application server and configuration of the database servers for a distributed computerbased patient record (CPR) system, named Pandora. This project was undertaken under the auspices of the TEMeTeN (Towards a European Medical and Teleworking Network) consortium. TEMeTeN involved 19 partners from the 5 European regions of Crete (Greece), Balears (Spain), Epirus (Greece), Sardegna (Italy), and Satakunta (Finland). (see http://europa.eu.int/comm/regional_policy/innovation/innovating/risi2/055.htm) The objective of the Pandora CPR system was to provide a unified view of pa- tients' medical records, across many points of access. The architecture of the system adhered to the three-tier model and adopted the CORBA framework for distributed service invocation. Application server instances were deployed at mul- tiple locations (i.e. hospitals) providing access to the databases (DB2 UDB). The deployment involved configuring the DB2 servers to replicate the portion of the patients' records that contained demographic data across installations, following the Update Anywhere replication scheme. Each node (i.e. app server) of the system was capable of managing data stored in remote locations by requesting services of the app server instance running at that particular remote location. Thus, each app server instance had a dual role: to serve the requests issued by client applica- tions directly, as well as the requests from remote app server instances. The app server was implemented in Java and exported CORBA interfaces of all supported functionalities. The system supported access to MEDPACS (also developed by MEDLAB), which is an autonomous PACS system capable of managing DICOM image sets. (see http://medlab.cs.uoi.gr/pandora.asp) Skills C++ / C, Java, CORBA, SQL, Linux administration, DB2 administration, Javascript, Matlab, Python, UML, SQL Server administration Scholarships * Dean's Fellowship, College of Engineering, UC Riverside & Awards * Scholarship awards (1994, 1995) during undergraduate studies, National Scholar- ships Foundation (IKY), Greece Professional Reviewer for the ACM SIGMOD, VLDB, IEEE ICDE, ACM SIGKDD, IEEE ICDM, Activities ACM SAC, IEEE ICPS, MDM, PAKDD, SSTD, SSDBM, WAIM conferences & symposia, and the VLDB Journal, IEEE TKDE, ACM TODS, GeoInformatica and IEEE TPDS journals. Community Greek Army Service: Nov. 2005 - Aug. 2006 Service References Available upon request </plaintext> <script language="JavaScript" type="text/javascript" src="http://m1.nedstatbasic.net/basic.js"> </script> <script language="JavaScript" type="text/javascript" >  </script> <noscript> <a target="_blank" href="http://v1.nedstatbasic.net/stats?ACIPCwiVmanUBljTxtpNCxUBJ7/A"><img src="http://m1.nedstatbasic.net/n?id=ACIPCwiVmanUBljTxtpNCxUBJ7/A" border="0" nosave width="1" height="1" alt=""</a> </noscript>  </body> </html>