PDM 2006: Proceedings of the International Workshop on Data Mining

Recently major processor manufacturers have announced a dramatic shift in their paradigm to increase computing power over the coming years. Instead of focusing on faster clock speeds and more powerful single core CPUs, the trend clearly goes towards multi core systems. This will also result in a paradigm shift for the development of algorithms for computationally expensive tasks, such as data mining applications. Obviously, work on parallel algorithms is not new per se but concentrated efforts in the many application domains are still missing. Multi-core systems, but also clusters of workstations and even large-scale distributed computing infrastructures provide new opportunities and pose new challenges for the design of parallel and distributed algorithms. Since data mining and machine learning systems rely on high performance computing systems, research on the corresponding algorithms must be on the forefront of parallel algorithm research in order to keep pushing data mining and machine learning applications to be more powerful and, especially for the former, interactive. To bring together researchers and practitioners working in this exciting field, a workshop on parallel data mining was organized as part of PKDD/ECML 2006 (Berlin, Germany). The six contributions selected for the program describe various aspects of data mining and machine learning approaches featuring low to high degrees of parallelism: The first contribution focuses the classic problem of distributed association rule mining and focuses on communication efficiency to improve the state of the art. After this a parallelization technique for speeding up decision tree construction by means of thread-level parallelism for shared memory systems is presented. The next paper discusses the design of a parallel approach for dis- tributed memory systems of the frequent subgraphs mining problem. This approach is based on a hierarchical communication topology to solve issues related to multi-domain computational envi- ronments. The forth paper describes the combined use and the customization of software packages to facilitate a top down parallelism in the tuning of Support Vector Machines (SVM) and the next contribution presents an interesting idea concerning parallel training of Conditional Random Fields (CRFs) and motivates their use in labeling sequential data. The last contribution finally focuses on very efficient feature selection. It describes a parallel algorithm for feature selection from random subsets. Selecting the papers included in this volume would not have been possible without the help of an international Program Committee that has provided detailed reviews for each paper. We would like to also thank Matthew Otey who helped with publicity for the workshop.

Dietary intakes and food sources of phytoestrogens in the European Prospective Investigation into Cancer and Nutrition (EPIC) 24-hour dietary recall cohort

BACKGROUND/OBJECTIVES: Phytoestrogens are estradiol-like natural compounds found in plants that have been associated with protective effects against chronic diseases, including some cancers, cardiovascular diseases and osteoporosis. The purpose of this study was to estimate the dietary intake of phytoestrogens, identify their food sources and their association with lifestyle factors in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. SUBJECTS/METHODS: Single 24-hour dietary recalls were collected from 36 037 individuals from 10 European countries, aged 35–74 years using a standardized computerized interview programe (EPIC-Soft). An ad hoc food composition database on phytoestrogens (isoflavones, lignans, coumestans, enterolignans and equol) was compiled using data from available databases, in order to obtain and describe phytoestrogen intakes and their food sources across 27 redefined EPIC centres. RESULTS: Mean total phytoestrogen intake was the highest in the UK health-conscious group (24.9 mg/day in men and 21.1 mg/day in women) whereas lowest in Greece (1.3 mg/day) in men and Spain-Granada (1.0 mg/day) in women. Northern European countries had higher intakes than southern countries. The main phytoestrogen contributors were isoflavones in both UK centres and lignans in the other EPIC cohorts. Age, body mass index, educational level, smoking status and physical activity were related to increased intakes of lignans, enterolignans and equol, but not to total phytoestrogen, isoflavone or coumestan intakes. In the UK cohorts, the major food sources of phytoestrogens were soy products. In the other EPIC cohorts the dietary sources were more distributed, among fruits, vegetables, soy products, cereal products, non-alcoholic and alcoholic beverages. CONCLUSIONS: There was a high variability in the dietary intake of total and phytoestrogen subclasses and their food sources across European regions.

Towards expressive rule induction on IP network event streams

In order to gain insights into events and issues that may cause errors and outages in parts of IP networks, intelligent methods that capture and express causal relationships online (in real-time) are needed. Whereas generalised rule induction has been explored for non-streaming data applications, its application and adaptation on streaming data is mostly undeveloped or based on periodic and ad-hoc training with batch algorithms. Some association rule mining approaches for streaming data do exist, however, they can only express binary causal relationships. This paper presents the ongoing work on Online Generalised Rule Induction (OGRI) in order to create expressive and adaptive rule sets real-time that can be applied to a broad range of applications, including network telemetry data streams.

A Packet/Frame sync detector based on statistical mode with application to wireless-USB

Wireless Personal Area Networks (WPANs) are offering high data rates suitable for interconnecting high bandwidth personal consumer devices (Wireless HD streaming, Wireless-USB and Bluetooth EDR). ECMA-368 is the Physical (PHY) and Media Access Control (MAC) backbone of many of these wireless devices. WPAN devices tend to operate in an ad-hoc based network and therefore it is important to successfully latch onto the network and become part of one of the available piconets. This paper presents a new algorithm for detecting the Packet/Fame Sync (PFS) signal in ECMA-368 to identify piconets and aid symbol timing. The algorithm is based on correlating the received PFS symbols with the expected locally stored symbols over the 24 or 12 PFS symbols, but selecting the likely TFC based on the highest statistical mode from the 24 or 12 best correlation results. The results are very favorable showing an improvement margin in the order of 11.5dB in reference sensitivity tests between the required performance using this algorithm and the performance of comparable systems.

Multicore-enabling the MPJ express messaging library

With the transition to multicore processors almost complete, the parallel processing community is seeking efficient ways to port legacy message passing applications on shared memory and multicore processors. MPJ Express is our reference implementation of Message Passing Interface (MPI)-like bindings for the Java language. Starting with the current release, the MPJ Express software can be configured in two modes: the multicore and the cluster mode. In the multicore mode, parallel Java applications execute on shared memory or multicore processors. In the cluster mode, Java applications parallelized using MPJ Express can be executed on distributed memory platforms like compute clusters and clouds. The multicore device has been implemented using Java threads in order to satisfy two main design goals of portability and performance. We also discuss the challenges of integrating the multicore device in the MPJ Express software. This turned out to be a challenging task because the parallel application executes in a single JVM in the multicore mode. On the contrary in the cluster mode, the parallel user application executes in multiple JVMs. Due to these inherent architectural differences between the two modes, the MPJ Express runtime is modified to ensure correct semantics of the parallel program. Towards the end, we compare performance of MPJ Express (multicore mode) with other C and Java message passing libraries---including mpiJava, MPJ/Ibis, MPICH2, MPJ Express (cluster mode)---on shared memory and multicore processors. We found out that MPJ Express performs signicantly better in the multicore mode than in the cluster mode. Not only this but the MPJ Express software also performs better in comparison to other Java messaging libraries including mpiJava and MPJ/Ibis when used in the multicore mode on shared memory or multicore processors. We also demonstrate effectiveness of the MPJ Express multicore device in Gadget-2, which is a massively parallel astrophysics N-body siimulation code.

Resilient responses to climate change: exploring organisational learning in networked development organisations

Drawing from the organisational learning and governance literature, this paper assesses four internationally networked governmental and non‐governmental organisations in the UK addressing climate change. We analyse how those concerned understand the climate change crisis, what mechanisms are put in place to address information flows, and what evidence there is of learning through sharing information between the organisational headquarters and their regional offices. The most striking finding is the evidence of learning that largely depends on ad‐hoc informal processes and shadow networks.

Quality time [QoS in 802.11e]

Wireless local area networks (WLANs) based on the IEEE 802.11 standard are now widespread. Most are used to provide access for mobile devices to a conventional wired infrastructure, and some are used where wires are not possible, forming an ad hoc network of their own. There are several varieties at the physical or radio layer (802.11, 802.11a, 802.11b, 802.11g), with each featuring different data rates, modulation schemes and transmission frequencies. However, all of them share a common medium access control (MAC) layer. As this is largely based on a contention approach, it does not allow prioritising of traffic or stations, so it cannot easily provide the quality of service (QoS) required by time-sensitive applications, such as voice or video transmission. In order to address this shortfall of the technology, the IEEE set up a task group that is aiming to enhance the MAC layer protocol so that it can provide QoS. The latest draft at the time of writing is Draft 11, dated October 2004. The article describes the yet-to-be-ratified 802.11e standard and is based on that draft.

A systematic approach to identify the sources of tropical SST errors in coupled models using the adjustment of initialised experiments

Understanding the sources of systematic errors in climate models is challenging because of coupled feedbacks and errors compensation. The developing seamless approach proposes that the identification and the correction of short term climate model errors have the potential to improve the modeled climate on longer time scales. In previous studies, initialised atmospheric simulations of a few days have been used to compare fast physics processes (convection, cloud processes) among models. The present study explores how initialised seasonal to decadal hindcasts (re-forecasts) relate transient week-to-month errors of the ocean and atmospheric components to the coupled model long-term pervasive SST errors. A protocol is designed to attribute the SST biases to the source processes. It includes five steps: (1) identify and describe biases in a coupled stabilized simulation, (2) determine the time scale of the advent of the bias and its propagation, (3) find the geographical origin of the bias, (4) evaluate the degree of coupling in the development of the bias, (5) find the field responsible for the bias. This strategy has been implemented with a set of experiments based on the initial adjustment of initialised simulations and exploring various degrees of coupling. In particular, hindcasts give the time scale of biases advent, regionally restored experiments show the geographical origin and ocean-only simulations isolate the field responsible for the bias and evaluate the degree of coupling in the bias development. This strategy is applied to four prominent SST biases of the IPSLCM5A-LR coupled model in the tropical Pacific, that are largely shared by other coupled models, including the Southeast Pacific warm bias and the equatorial cold tongue bias. Using the proposed protocol, we demonstrate that the East Pacific warm bias appears in a few months and is caused by a lack of upwelling due to too weak meridional coastal winds off Peru. The cold equatorial bias, which surprisingly takes 30 years to develop, is the result of an equatorward advection of midlatitude cold SST errors. Despite large development efforts, the current generation of coupled models shows only little improvement. The strategy proposed in this study is a further step to move from the current random ad hoc approach, to a bias-targeted, priority setting, systematic model development approach.