Back to Top

Publications

■ Corridor Learning Using Individual Trajectories
Nikolaos Zygouras, Dimitrios Gunopulos: Corridor Learning Using Individual Trajectories. In Mobile Data Management (MDM), 2018 19th IEEE International Conference on. IEEE. (to appear)
In this work, we proposed a pipelined approach for detecting a set of frequently accessed corridors from a vast collection of trajectories. Initially we applied a well known topic modelling technique to detect frequent sets of locations and then we derived the frequent corridors from the trajectories that accessed these locations.
■ Urban Travel Time Prediction using a Small Number of GPS-floating Cars
Li, Y., Gunopulos, D., Lu, C., & Guibas, L. (2017, November). Urban Travel Time Prediction using a Small Number of GPS Floating Cars. In Proceedings of the 25th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (p. 3). ACM.
While real-time collection of GPS trajectories from taxis and mobile users are common today, a practical solution for trajectory-based travel time prediction needs to be robust to the situation of only having access to a small number of active mobile probes. This paper presented an algorithm framework for predicting path travel time from GPS trajectories, under the scenario of 10-15 of GPS-floating cars and no trip labels.
Get the paper
■ Discovering Corridors From GPS Trajectories
Zygouras, N., & Gunopulos, D. (2017, November). Discovering Corridors From GPS Trajectories. In Proceedings of the 25th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (p. 61). ACM.
In this work a pipelined approach is proposed for detecting a set of frequently accessed paths, named as corridors, from a vast collection of trajectories. Initially we applied a well known topic modelling technique to detect frequent sets of locations and then we derived frequent paths at these locations. Our initial experimental results demonstrate the ability of our approach to summarize a large collection of trajectories to a few number of frequently accessed paths. The detection of such corridors abstracts the complex trajectories and returns the major movement patterns.
Get the paper
A Cognitive Map-Based Representation for Consumer Behaviour Modelling
Homenda, W., & Jastrzebska, A. (2017, November). A Cognitive Map-Based Representation for Consumer Behaviour Modelling. In the Eleventh International Conference on Advances in Semantic Processing (pp. 1-7). IARIA.
In many areas of science, there is a need for modelling approaches that represent knowledge in an abstract, generalized way. This need manifests itself mainly, when models are designed to be interpreted and used by human beings, but not exclusively. Abstraction comes in hand, when we deal with very large data sets. When standard numerical methods are inadequate to describe the data, it might be beneficial to turn towards granular models, based on concepts, where knowledge is aggregated and represented in an abstract fashion. Concepts-based methods facilitate smooth human-computer interactions as they allow to represent knowledge in form of relationships between phenomena – just like humans do.
Get the paper
■ Resource Allocation with Population Dynamics
Epperlein, J. and Marecek, J., "Resource Allocation with Population Dynamics", 55th Annual Allerton Conference on Communication, Control, and Computing, 2017.
The model of Marecek et al. [arXiv:1406.7639, Int. J. Control 88(10), 2015] is extended to consider a model of the evolution of a heterogeneous population of agents over time, governed by a Markov chain. Still, we are able to show that the distribution of agents across resources converges in distribution, for suitable means of information provision, under certain assumptions.
Get the paper
■ On Classical Control and Smart Cities
Fioravanti, Andre R and Marecek, Jakub and Shorten, Robert N and Souza, Matheus and Wirth, Fabian R, "On Classical Control and Smart Cities", 56th IEEE Conference on Decision and Control, 2017.
By means of an example and associated theory we show that many classical controllers are not suitable for deployment in smart city applications. We use tools from iterated function systems to identify controllers that can be used to design not only stable closed-loop systems, but also to regulate large-scale populations of agents in a predictable manner.
Get the paper
■ Estimation of Delays for Individual Trams to Monitor Issues in Public Transport Infrastructure
Luckner, M., & Karwowski, J. (2017, September). Estimation of Delays for Individual Trams to Monitor Issues in Public Transport Infrastructure. In Conference on Computational Collective Intelligence Technologies and Applications (pp. 518-527). Springer, Cham. ISO 690
Open stream data on public transport published by cities can be used by third party developers such as Google to create a real-time travel planner. However, such system should predict future events on roads. We have used open stream data with current trams' localisations and timetables to estimate current delays of individual trams. On that base, we calculate a global coefficient that can be used as a measure to monitor a current situation in a public transport network. Fig. 1 shows how the ratio of the trams with huge delay exceeds the ratio of the tram with low delay. That is an impulse to predict future problems with public transport. Fig. 2 compares the distribution of delayed trams when the alarm was raised and one hour later. That shows how a critical situation for a public transport network can be detected before the peak points of cumulative delays.
Get the paper
■ An Interactive Approach for Exploration of Flows through Direction-Based Filtering
Katerina Vrotsou, Georg Fuchs, Natalia Andrienko, and Gennady Andrienko, "An Interactive Approach for Exploration of Flows through Direction-Based Filtering", Journal of Geovisualization and Spatial Analysis, 2017, vol. 1(1), (accepted)
This paper proposes an alternative approach to visual analysis of origin-destination data. The approach is based on a flow-specific interaction technique for filtering the data by direction that enables an analyst to successively identify underlying spatial arrangement patterns.
Get the paper
■ One for All and All for One: Simultaneous Approximation of Multiple Functions over Distributed Streams
Lazerson, A., Gabel, M., Keren, D., & Schuster, A. (2017, June). One for All and All for One: Simultaneous Approximation of Multiple Functions over Distributed Streams. In Proceedings of the 11th ACM International Conference on Distributed and Event-based Systems (pp. 203-214). ACM.

Get the paper
■ A Pareto-based scheduler for exploring cost-performance trade-offs for MapReduce workloads
A Pareto-based scheduler for exploring cost-performance trade-offs for MapReduce workloads, Nikos Zacheilas and Vana Kalogeraki, EURASIP Journal on Embedded Systems, Springer July 3, 2017

Get the paper
■ Maximizing Determinism in Stream Processing Under Latency Constraints
Maximizing Determinism in Stream Processing Under Latency Constraints, Nikos Zacheilas, Vana Kalogeraki, Yiannis Nikolakopoulos, Vincenzo Gulisano, Marina Papatriantafilou, Philippas Tsigas, ACM DEBS, Barcelona, Spain, June 19 - 23, 2017 (received best paper award)

Get the paper
■ DIsCO: DynamIc Data COmpression in Distributed Stream Processing Systems
DIsCO: DynamIc Data COmpression in Distributed Stream Processing Systems, Zacheilas Nikos, Vana Kalogeraki, DAIS 2017, Neuchâtel, Switzerland, June 19 - 22, 2017

Get the paper
■ A Framework for Efficient Energy Scheduling of Spark Workloads
A Framework for Efficient Energy Scheduling of Spark Workloads, Stathis Maroulis, Nikos Zacheilas, Vana Kalogeraki , IEEE ICDCS 2017, Atlanta, GA, USA, June 5 - 8, 2017 (poster)

Get the paper
■ Pythia: A System for Online Topic Discovery of Social Media Posts
Pythia: A System for Online Topic Discovery of Social Media Posts, Iouliana Litou, Vana Kalogeraki, IEEE ICDCS 2017, Atlanta, GA, USA, June 5 - 8, 2017 (Demo)

Get the paper
■ Influence Maximization in a Many Cascades World
Influence Maximization in a Many Cascades World, Iouliana Litou, Vana Kalogeraki, Dimitrios Gunopulos, IEEE ICDCS, Atlanta, GA, USA, June 5 - 8, 2017

Get the paper
■ Lessons Learnt from the analysis of a bike sharing system
Lessons Learnt from the analysis of a bike sharing system, Dimitrios Tomaras, Ioannis Boutsis, Vana Kalogeraki, 10th International PETRA 2017, Rhodes, Greece, June 21 - 23, 2017

Get the paper
■ Dynamic map update of non-static facility logistics environment with a multi-robot system
N. Shaik, T. Liebig, C. Kirsch, and H. Müller, “Dynamic map update of non-static facility logistics environment with a multi-robot system,” in Proceedings of the 40th German Conference on Artificial Intelligence, Springer Berlin Heidelberg, 2017, p. (accepted).

Get the paper
■ Real-time Public Transport Delay Prediction for Situation-aware Routing
L. Heppe and T. Liebig, “Real-time Public Transport Delay Prediction for Situation-aware Routing,” in Proceedings of the 40th German Conference on Artificial Intelligence, Springer Berlin Heidelberg, 2017, p. (accepted).

Get the paper
■ Combining Stream Mining and Neural Networks for Short Term Delay Prediction
M. Grzenda, K. Kwasiborska, T. Zaremba: Combining Stream Mining and Neural Networks for Short Term Delay Prediction. Accepted in International Conference on Soft Computing Models in Industrial and Environmental Applications, Leon, Spain. SOCO’17. 6th-8th September 2017
The location readings from GPS-based devices in public transport vehicles are received with latency caused by partly irregular data transmission; such latency hinders real-time identification of delays. Fig. 1 shows the histogram of sample latencies in location data acquisition. In this work we proposed a hybrid method combining stream mining (Hoeffding tree) and batch learning (multilayer perceptron) increasing the accuracy of the prediction of tram delay statuses. Such predicted statuses can be used as a temporary replacement of ground truth tram delay statuses, not available in real-time due to the aforementioned data transmission latencies.
Get the paper
■ REMI, Reusable Elements for Multi-Level Information Availability
A. Senderovich, N. Rivetti, A. Gal, N. Panagiotou, I. Katakis, D. Gunopulos, V. Kalogeraki: REMI, Reusable Elements for Multi-Level Information Availability. Accepted in DEBS 2017
We provide a framework that enables data solutions to run on less-than-perfect sources of information. We name our solution REMI for Reusable Elements for Missing Information. Our approach comprises of elements that can adapt themselves, by design, to the changing levels of data availability. The core of the REMI framework is the layered stack architecture, equipped with data processing units with two additional computational elements, namely DARE (data enrichment) and GRADE (graceful degradation).
Get the paper

Pages