@article {18510, title = {One-class classifiers based on entropic spanning graphs}, journal = {IEEE Transactions on Neural Networks and Learning Systems}, volume = {28}, issue = {12}, year = {2017}, month = {11/2016}, pages = { 2846 - 2858}, chapter = {2846}, abstract = {One-class classifiers offer valuable tools to assess the presence of outliers in data. In this paper, we propose a design methodology for one-class classifiers based on entropic spanning graphs. The spanning graph is learned on the embedded input data, with the aim to generate a partition of the vertices. The final partition is derived by exploiting a criterion based on mutual information minimization. Here, we compute the mutual information by using a convenient formulation provided in terms of the alfa-Jensen difference. Once training is completed, in order to associate a confidence level with the classifier decision, a graph-based fuzzy model is constructed. The fuzzification process is based only on topological information of the vertices of the entropic spanning graph. As such, the proposed one-class classifier is suitable also for datasets with complex geometric structures. We provide experiments on well-known benchmarking datasets containing both feature vectors and labeled graphs. In addition, we apply the method on the problem of protein solubility recognition by considering several data representations for the samples. Experimental results demonstrate the effectiveness and versatility of the proposed method with respect to other state-of the-art approaches. }, doi = { 10.1109/TNNLS.2016.2608983}, author = {Livi, Lorenzo and Alippi, Cesare} } @conference {18448, title = {One-Class Classification Through Mutual Information Minimization}, booktitle = {IEEE-INNS International Joint Conference on Neural Networks (IJCNN16)}, year = {2016}, month = {07/2016}, address = {Vancouver, Canada}, abstract = {In one-class classification problems, a model is synthesized by using only information coming from the nominal state of the data generating process. Many important applications can be cast in the one-class classification framework, such as anomaly and change in stationarity detection, and fault recognition. In this paper, we present a novel design methodology for oneclass classifiers derived from graph-based entropy estimators. The entropic graph is used to generate a partition of the input nominal conditions, which corresponds to the classifier model. Here we propose a criterion based on mutual information minimization to learn such a partition. The _-Jensen difference is considered, which provides a convenient way for estimating the mutual information. The classifier incorporates also a fuzzy model, providing a confidence value for a generic test sample during operational modality, expressed as a membership degree of the sample to the nominal conditions class. The fuzzification mechanism is based only on topological properties of the entropic spanning graph vertices; as such, it allows to model clusters of arbitrary shapes. We show preliminary {\textendash} yet very promising {\textendash} results on both synthetic problems and real-world datasets for one-class classification. }, author = {Livi, Lorenzo and Alippi, Cesare} } @conference {18450, title = {Online Model-free Sensor Fault Identification and Dictionary Learning in Cyber-Physical Systems}, booktitle = {IEEE-INNS International Joint Conference on Neural Networks (IJCNN16)}, year = {2016}, month = {07/2016}, address = {Vancouver, Canada}, abstract = {This paper presents a model-free method for the online identification of sensor faults and learning of their fault dictionary. The method, designed having in mind Cyber-Physical Systems (CPSs), takes advantage of functional relationships among the datastreams acquired by CPS sensing units. Existing model-free change detection mechanisms are proposed to detect faults and identify the fault type thanks to a fault dictionary which is built over time. The main features of the proposed algorithm are its ability to operate without requiring any a priori information about the system under inspection or the nature of the possibly occurring faults. As such, the method follows the model-free approach, characterized by the fact the fault dictionary is constructed online once faults are detected. Whenever available, humans can be considered in the loop to label a fault or a fault class in the dictionary as well as introduce fault instances generated thanks to a priori information. Experimental results on both synthetic and real datasets corroborate the effectiveness of the proposed fault diagnosis system. }, author = {Alippi, Cesare and Ntalampiras, Stavros and Roveri, Manuel} } @conference {18454, title = {Optimizing Failure Prediction to Maximize Availability}, booktitle = {13th IEEE International Conference on Autonomic Computing (ICAC)}, year = {2016}, month = {07/2016}, address = {W{\"u}rzburg, Germany}, abstract = {Availability of autonomous systems can be enhanced with self-monitoring and fault-tolerance methods based on failures prediction. With each correct prediction, proactive actions may be taken to prevent or to mitigate a failure. On the other hand, incorrect predictions will introduce additional downtime associated with the overhead of a proactive action that may decrease availability. The total effect on availability will depend on the quality of prediction (measured with precision and recall), the overhead of proactive actions (penalty), and the benefit of proactive actions when prediction is correct (reward). In this paper, we quantify the impact of failure prediction and proactive actions on steady-state availability. Furthermore, we provide guidelines for optimizing failure prediction to maximize availability by selecting a proper precision and recall trade-off with respect to penalty and reward. A case study to demonstrate the approach is also presented.}, author = {Kaitovi{\'c}, Igor and Malek, Miroslaw} } @conference {18380, title = {Optimizing Sensor Nodes Placement for Fault-tolerant Trilateration-based Localization}, booktitle = {IEEE Pacific Rim International Symposium on Dependable Computing (PRDC)}, year = {2015}, month = {11/2015}, address = {Zhangjiajie, China}, author = {Bala{\'c}, Katarina and Prevostini, Mauro and Malek, Miroslaw} } @article {156.MaPaSiZa12.TCAD, title = {OSCAR: an Optimization Methodology Exploiting Spatial Correlation in Multi-core Design Space}, journal = {IEEE Transactions on Computer-Aided Design}, volume = {21}, number = {-}, issue = {5}, year = {2012}, month = {05/2012}, pages = {740-753}, publisher = {IEEE}, abstract = {This paper presents OSCAR, an Optimization methodology exploiting Spatial CorrelAtion of multi-coRe design space. The paper builds upon the observation that power consumption and performance metrics of spatially close design configurations (or points) are statistically correlated. We propose to exploit the correlation by using a Response Surface Model (RSM), i.e., a closed-form expression suitable for predicting the quality of non-simulated design points. This model is useful during the design space exploration (DSE) phase to quickly converge to the Pareto set of the multi-objective problem without executing lengthy simulations. We compare the proposed heuristic with state-of-the-art approaches (conventional, RSM-based and structured DOEs). Experimental results show that OSCAR is a faster heuristic with respect to state of the art techniques such as Response-Surface Pareto Iterative Refinement - ReSPIR and Nondominated Sorting Genetic Algorithm - NSGA-II. Reported results also show that OSCAR can significantly improve structured DOE approaches by slightly increasing the number of experiments.}, keywords = {chip multi processor, correlation based design, design space exploration, multi-core, multi-objective optimization, OSCAR}, author = {Mariani, Giovanni and Palermo, Gianluca and Silvano, Cristina and Zaccaria, Vittorio} } @conference {148.DeKaFi11.NOCS, title = {Online Task Remapping Strategies for Fault-tolerant Network-on-Chip Multiprocessors}, booktitle = {NOCS {\textquoteright}11: Proceedings of the Fifth ACM/IEEE International Symposium on Networks-on-Chip}, year = {2011}, month = {05/2011}, pages = {1{\textendash}8}, address = {Pittsburgh, Pennsylvania, USA}, abstract = {As CMOS technology scales down into the deep submicron domain, the aspects of fault tolerance in complex Networks-on-Chip (NoCs) architectures are assuming an increasing relevance. Task remapping is a software based solution for dealing with permanent failures in processing elements in the NoC. In this work, we formulate the optimal task mapping problem for mesh-based NoC multiprocessors with deterministic routing as an integer linear programming (ILP) problem with the objective of minimizing the communication traffic in the system and the total execution time of the application. We find the optimal mappings at design time for all scenarios where single-faults occur in the processing nodes. We propose heuristics for the online task remapping problem and compare their performances with the optimal solutions.}, keywords = {adaptivity, fault tolerance, kahn process networks (KPN), mapping, network-on-chip (NoC), self-adaptivity}, doi = {http://dx.doi.org/10.1145/1999946.1999967}, author = {Derin, Onur and Kabakci, Deniz and Fiorin, Leandro} } @inbook {143.RiKaTuPaSiZaMa.2011, title = {Optimization Algorithms for Embedded System Design Space Exploration}, booktitle = {Multi-objective design space exploration of multiprocessor SoC architectures: the MULTICUBE approach}, year = {2011}, publisher = {Springer}, organization = {Springer}, address = {New York, USA}, abstract = {This paper is dedicated to the optimization algorithms developed in the MULTICUBE project and to their surrounding environment. Two software design space exploration (DSE) tools host the algorithms: Multicube Explorer and mode-FRONTIER. The description of the proposed algorithms is the central part of the paper. The focus will be on newly developed algorithms and on ad-hoc extensions of existing techniques in order to face with discrete and categorical design space parameters that are very common when working with embedded systems design. This paper will also provide some fundamental guidelines to build a strategy for testing the performance and accuracy of such algorithms. The aim is mainly to build confidence in optimization techniques, rather than to simply compare one algorithm versus another one. The no-free-lunch theorem for optimization has to be taken into consideration and therefore the analysis will look forward to robustness and industrial reliability of the results.}, author = {Rigoni, Enrico and Kavka, Carlos and Turco, Alessandro and Palermo, Gianluca and Silvano, Cristina and Zaccaria, Vittorio and Mariani, Giovanni} }