Tutorial at the 2017 International Joint Conference on Neural Networks
(IJCNN 2017) Anchorage, Alaska, USA, May 14-19, 2017
http://www.ijcnn.org/
Tutorial title: Time-Evolving Data Streams Learning and Short-Term Urban
Traffic Flow Forecasting
http://www.disi.unige.it/person/MasulliF/ricerca/IJCNN2017/
Date: Sunday May 14th, 2017; Time 1:30 pm - 3:30 pm
Presenter: Prof. Francesco Masulli
DIBRIS - Dept of Informatics, Bioingengering, Robotics and Systems
Engineering, University of Genova (ITALY)
and Center for Biotechnology of Temple University, Philadelphia (PA, USA)
Abstract: Data streams have arisen as a relevant topic during the last
decade. In this tutorial we consider non-stationary data stream
clustering using a possibilistic approach. The Graded Possibilistic
Clustering model offers a way to evaluate “outlierness” through a
natural measure, which is computed directly from the model. Both online
and batch training approaches are considered, to provide different
trade-offs between stability and speed of response to changes. The
proposed approach is evaluated on a synthetic data set, for which the
ground truth is available. Moreover, a real-time short-term urban
traffic flow forecasting application is proposed, taking into account
both spatial and temporal information. To this aim, we introduce a
Layered Ensemble Model which combines Artificial Neural Networks and
Graded Possibilistic Clustering models, obtaining in such a way an
accurate forecaster of the traffic flow rates with outlier detection.
Experimentation has been carried out on two different data sets. The
former consists on real UK motorway data and the latter is obtained from
simulated traffic flow on a street network in Genoa (Italy). The
proposed model for short-term traffic forecasting provides promising
results and given its characteristics of outlier detection, accuracy,
and robustness, and can be fruitful integrated in traffic flow
management systems.
