LEAPS 2013 - 1st International Workshop on Learning stratEgies and dAta Processing in nonStationary environments

Most machine learning techniques assume, either explicitly or implicitly, that the data-generating process is stationary. This assumption guarantees that the model learnt during the initial training phase remains valid over time and that its performance is in line with our expectations. Unfortunately, this assumption does not truly hold in the real world representing, in many cases, a simplistic approximation of the reality.
Data from real-world scenarios are often affected by nonstationarities and, during operational life, their describing model (or distribution) diverges from the one that yielded the training set. Among the causes generating nonstationarity we mention natural (and unknown) evolutions of the data-generating process, faults/aging affecting sensing and processing devices and model bias introduced by a poor training set. Learning-based systems have to be up-to-date to be effective, thereby requiring adaptation mechanisms to deal with nonstationary environments.
In machine learning nonstationarity is referred to as concept drift and several techniques to detect and adapt to concept drift have been presented in different application domains e.g., fraud detection in electronic transactions, sensor networks, intelligent vehicles and recommender systems. Other relevant scenarios are classification systems designed to cope with concept drift, such as those addressing email/spam filtering, internet events log analysis, stock market forecasting, context-aware and ubiquitous computing.
The workshop focuses on intelligent solutions to analyze/process data acquired in nonstationary environments. Original contributions in the field of fault detection and diagnosis, as well as cognitive approaches for learning characteristics of the process to handle nonstationarity are particularly welcome.
We encourage submissions presenting novel theoretical, methodological or experimental results.
Topics of interest
The workshop topics include but are not limited to, ethical, social and philosophical issues of Artificial Intelligence Applications relating to:
Computational Intelligent solution for Fault Detection/ Isolation/ Identification
Change-Detection Tests (or Novelty-Detection Tests)
Change Detection exploiting contextual information
Adaptive Classifiers for Concept Drift
Concepts Drift and Recurring Concept management
Embedded systems implementing computational intelligence techniques to achieve intelligent behavior in nonstationary environments
Adaptive solutions to operate in evolving/faulty environments
Intelligent embedded systems for applications such as:
intelligent buildings
robotics
homeland security
environmental monitoring
sensor networks
water distribution networks
Intrusion detection in computer networks
Application domains where data are affected by concept drift
Workshop Chairs
Giacomo Boracchi, Politecnico di Milano, Italy giacomo.boracchi-AT-polimi.it
Manuel Roveri, Politecnico di Milano, Italy manuel.roveri-AT-polimi.it
Technical Program Committee
Rami Abielmona, University of Ottawa, Canada
Haibo He, University of Rhode Island, US
Vasso Reppa, KIOS Research Center for Intelligent Systems and Networks, Cyprus
Michalis P. Michaelides, Cyprus University of Technlolgy
Stefano Zanero, Politecnico di Milano, Italy
Peter Tino, University of Birmingham, UK
Vicenç Puig, Universitat Politècnica de Catalunya, Spain
Maurizio Bocca, University of Utah, USA
Vincent Lemaire, Orange Labs, France
Alessandro Lazaric, INRIA Lille, France
Daniele Caltabiano, STMicroelectronics, Milano Italy
Leandro L. Minku, The University of Birmingham, UK