Vai al contenuto| Home page|

   Ti trovi in: HOME »Programmi, progetti e risultati »I progetti »PRIN - Programmi di ricerca di Rilevante Interesse Nazionale»Programma di ricerca
INIZIO_TESTO_DA_INDICIZZARE

RESEARCH PROGRAM

italiano - inglese
Research Units
Similar research programs:
Scientific and education field classification
International Patent Classification
Bibliografia
[1] Chomicki J. and Toman D. (Eds.) (2005). Proc. 12th Int. Symp. on Temporal Representation and Reasoning (Burlington, Vermont, USA).

[2] Cohn A. nd Mrk D. (Eds.) (2005). Proc. Int. Conf. on Spatial Information Theory (Ellicottville, NY, USA).

[3] Rabiner L. (1989). "A tutorial on hidden markov models and selected applications in speech recognition". Proceedings of IEEE, 77, 57-286.

[4] Regazzoni C. and De Natale F. (Eds.) (2005). Proc. IEEE Int. Conf. on Image Processing (Genoa, Italy).

[5] Andrienko G., Malerba D., May M., and Teisseire M. (Eds.) (2005). Proc. ECML/PKDD Workshop on Mining Spatio-Temporal Data (Porto, Portugal).

[6] Durbin R., Eddy S., Krogh A., and Mitchison G. (1998). Biological sequence analysis, Cambridge Univ. Press, UK.

[7] Valdes A. (Ed.) (2005). Proc. Int. Symp. on Recent Advances in Intrusion Detection (Seattle, Washington, USA).

[8] Shahabi C. and Boucelma O. (Eds.) (2005). Proc. 13th Int. Symp. of ACM GIS (Bremen, Germany).

[9] Bayardo R. and Bennett K. (Eds.) (2005). Proc. 1th ACM Int. Conf. on Knowledge Discovery and Data Mining (Chicago, IL, USA).

[10] Muggleton S. and De Raedt L. (1994). “Inductive logic programming: theory and methods”. J. of Logic Programming, 19/20, 629–679.

[11] Levinson S. (1986). "Continuous variable duration hidden markov models for automatic speech recognition". Computer Speech and Language, 1, 29-45.

[12] Bonafonte A.B.J. (1996). “Duration modeling with expanded hmm applied to speech recognition”. citeseer.ist.psu.edu/507067.html

[13] Fine S., Singer Y., and Tishby N. (1998). "The hierarchical hidden markov model: Analysis and applications". Machine Learning, 32, 41-62.

[14] Murphy K. and Paskin M. (2001). "Linear time inference in hierarchical HMMs". Advances in NIPS, vol. 14.

[15] Xie L., Chang S., Divakaran A., and Sun H. (2002). "Learning hierarchical hidden Markov models for video structure discovery", T.R. 2002-006. ADVENT Group, Columbia Univ., New York, NY.

[16] Galassi U., Giordana A., Botta M. (2006). "Hierarchical Hidden Markov Models for Profile Learning". Fundamenta Informaticae. To appear.

[17] Li J., Gray R.M., Olshen R. A. (2000. "Multiresolution image classification by hierarchical modeling with two dimensional hidden Markov models". IEEE Trans. on Information Theory, 46, 1826-41.

[18] Joshi D., Li J. and Wang J. Z. (2006). "A Computationally Efficient Approach to the Estimation of Two- and Threedimensional Hidden Markov Models,'' IEEE Trans. on Image Processing. To appear.

[19] Feldman D. and Crutchfield J. (1998). "Measures of Statistical Complexity: Why?". Phys. Lett. A, 238, 244-252.

[20] Li W, (1991). "On the Relationship Between Complexity and Entropy for Markov Chains and Regular Languages". Complex Systems, 5, 381-399.

[21] W. Hamscher, L. Console and J. de Kleer (Eds.) (1992). Readings in Model-Based Diagnosis. Morgan Kaufmann.

[22] M. Sampath, R. Sengupta, S. Lafortune, K. Sinnamohideen, and D. Teneketzis. (1995). “Diagnosability of discrete event system”. IEEE Trans. on Automatic Control, 40, 1555-1575.

[23] L.Console, C. Picardi, D.Theseider Dupré (2003). “Temporal Decision Trees: Model-based Diagnosis of Dynamic Systems On-Board”. J. Artif. Intell. Res., 19, pp. 469-512.

[24] M.Botta, A.Giordana, L.Saitta and M.Sebag (2003). “Relational Learning as Search in a Critical Region”. Int. J. of Machine Learning Research, 4, 431-463.

[25] H. Ade, L. D. Raedt, and M. Bruynooghe (1995). “Declarative bias for specific to-general ILP systems”. Machine Learning, 20, 19–154.

[26] F. Zelezny, A. Srinivasan, D. Page (2004). “A Monte Carlo Study of Randomised Restarted Search in ILP”, In Proc. ILP-04, pp. 341-358.

[27] J.H. Gennari, P. Langley, and D. Fisher (1989). “Model of incremental concept formation”. Artificial Intelligence, 40, 11-61.

[28] S. Wrobel (1994). Concept Formation and Knowledge Revision, Kluwer Academic Publ.

[29] F. Richter (Ed.) (2006). In Order to Learn: How ordering processes and sequencing effects in machines illuminate human learning and vice-versa. Cambridge Univ. Press. To appear.

[30] Cornuéjols A. (2006). "Machine Learning : The necessity of order". In In Order to Learn: How ordering processes and sequencing effects in machines illuminate human learning and vice-versa, F. Richter. (Ed.), Cambridge University Press. To appear.

[31] Holte R. and Choueiry B. (2003). "Abstraction and reformulation in artificial intelligence". Phil. Trans. of the Royal Society - B, vol. 358, 1197 - 1204.

[32] Saitta L. (Ed.) (2003). The Abstraction Path, Special Issue of the Phil. Trans. of the Royal Soc. - Series B, vol. 358.

[33] Giunchiglia F., and Walsh T. (1992). "A Theory of Abstraction". Artificial Intelligence, 56, 323-390.

[34] Nayak P.P., and Levy A.Y. (1995). "A Semantic Theory of Abstraction". In Proc. IJCAI-95 (Montréal, Canada), pp. 196-202.

[35] Struss P. (1997). "Model-Based and Qualitative Reasoning: An Introduction". Ann. Math. Artif. Intell., 19, 355-381.

[36] Zucker J-D. (2003). "A grounded theory of abstraction in artificial intelligence", Phil. Trans. of the Royal Society - B, vol. 358, 1293 - 1309.

[37] Wneck J., and Michalski R. (1994). "Hypothesis-Driven Constructive Induction in AQ17-HCI: A Method and Experiments". Machine Learning, 14, 139-168.

[38] costruct induc recente

[39] Muggleton S., and Buntine W. (1988). "Machine Invention of First-Order Predicates by Inverting Resolution", In Proc. 5th Int. Conf. on Machine Learning (Ann Arbor, MI), pp. 339-352.

[40] I. Mozetic (1991). “Hierarchical Model-Based Diagnosis”. Int. J. of Man-Machine Studies, 35, 329-362.

[41] L. Console and D. Theseider Dupré (1994). “Abductive Reasoning with Abstraction Axioms”. LNCS, vol. 810, 98-112.

[42] K. Autio and R Reiter (1998). “Structural Abstraction in Model-Based Diagnosis”. In Proc. European Conf. on Artificial Intelligence, pp. 269-273.

[43] G. Provan (2001). “Hierarchical Model-Based Diagnosis”. In Proc. Int. Workshop on Principles of Diagnosis.

[44] L. Chittaro and R. Ranon (2004). “Hierarchical model-based diagnosis based on structural abstraction”. Artificial Intelligence, 155, 147-182.

[45] M. Sachenbacher, P. Struss (2005). “Task-dependent qualitative domain abstraction”. Artificial Intelligence, 162, 121-143.

[46] G. Torta and P. Torasso (2003). “Automatic Abstraction in Component-Based Diagnosis Driven by System Observability”. In Proc. Int. Joint Conference on Artificial Intelligence (Acapulco, Mexico), pp. 394-402.

[47] P.Torasso, G. Torta (2005). “Automatic Abstraction of Time-Varying System Models for Model Based Diagnosis”, LNAI, vol. 3698, 176-190.

[48] S. Ferilli, T. M. A. Basile, N. Di Mauro, and F. Esposito (2005). “On the Learnability of Abstraction Theories from Observations for Relational Learning”. LNAI, vol. 3720, 120-132.

[49] Saitta L. and Zucker J.D. (2000). "Abstraction and Phase Transition". In Proc. of the Int. Workshop on Abstraction, Reformulation and Approximation (Horseshoe Bay, TX), pp. 132-143.

[50] M. Gaines, W. Lisowski, S. Press, and N. Shapiro (1980). “Authentication by Keystroke Timing: some preliminary results”. Report R-256-NSF. Rand Corporation.

[51] M. Brown and S. Rogers (1993). “User identification via keystroke characteristics of typed names using neural networks”. Int. J. of Man-Machine Studies, 39, 999-1014.

[52] M. Obaidat and B. Sadoun (1997). “Verification of computer users using keystroke dynamics”. IEEE Trans. on Systems, Man, and Cybernetics - B, 27, 261-269.

[53] P. Dowland, and S. Furnell (2004). “A Long-term Trial of Keystroke Profiling using Digraph, Trigraph and Keyword Latencies”. In Proc. 19th Int. Conf. on Information Security (Toulouse, France).

[54] D. Gunetti and C. Picardi (2005). “Keystroke Analysis of free Text”. ACM Trans. on Information and System Security, 8, 312-347.
Keywords
MACHINE LEARNING, TEMPORAL/SPATIAL DATA, ABSTRACTION, HIDDEN MARKOV MODELS, MODEL-BASED DIAGNOSIS, INDUCTIVE LOGIC PROGRAMMING, INCREMENTAL LEARNING, HIERARCHICAL MODELS, 2D MULTIRESOLUTION HMM

Learning Hierarchical, Abstract Models from Temporal or Spatial Data

Abstract
The project is situated in the field of Machine Learning and Data Mining, specifically mining temporal or (one-dimensional or 2-dimensional) spatial data, with the possible help of available domain theories. The learning process is aimed at acquiring knowledge to be used in classification-related tasks (classification, diagnosis, monitoring, indexing, categorization, ...).

The overall goal of the project is to design a methodology, specifically targeting spatio/temporal data, that learns and uses knowledge fast, i.e., by keeping the computational requirements as low as possible, in order to allow scalability of both learning and performance systems well beyond the current frontiers.

To achieve this goal, a powerful and general tool has been identified in "abstraction", i.e., the mapping of data and knowledge from the original representation space to a "simpler" one, where problems are possibly easier to solve.

A substantial theoretical and experimental evidence indicates that a "good" abstraction allows problems to be solved with a computational complexity of orders of magnitude less, in the abstract space, than in the original one. However, the same evidence tells that a "good" abstraction is by no means easy to discover. Hence, the project, in order to maintain feasibility, shall limit the notion of abstraction to two special cases, that proved to be not only sufficiently powerful to cover >>>

Principal Investigator
Lorenza Saitta Università degli Studi del PIEMONTE ORIENTALE "Amedeo Avogadro"-Vercelli
Research Objectives
The project is situated in the field of Data Mining and Knowledge Discovery from temporally/spatially structured data and possibly background knowledge.

The activity of building models from data and/or discovering regularities is driven essentially by two motivations: acquisition of new knowledge, per se, and construction of systems devoted to specific tasks. This project essentially relies on the second motivation; in particular, the task aimed at is classification in a broad sense, including labelling, diagnosis, monitoring, identification, and recognition.

Starting from this perspective, the project targets three main goals:

Goal 1 - Automating the process of extracting knowledge and/or building models of 1D or 2D data (temporally or spatially ordered), possibly with the help of pre-existing knowledge, by using Machine Learning techniques.

Goal 2 - Increasing the scalability of both the knowledge learning step and the performance step using hierarchies, abstraction and background knowledge to cope with the restrictions imposed by limited resources.

Goal 3 - Comparing alternative approaches of model building in search of commonalities, possibly devising a generalized methodology, applicable across different domains.

Many real-world applications generate large amounts of data internally structured according to a temporal and/or spatial ordering. Examples include alarms/events >>>

Timescale
24 months
National and international background
Dealing with time and space is not new in Artificial Intelligence (AI). Since long time researchers have developed various types of temporal and spatial logics [1,2]. However, those approaches were meant to address theoretical issues concerning reasoning rather than to deal with large masses of data. In Pattern Recognition, also, analysis of time series or signal sequences (like speech) [3], as well as image processing [4] have reached a remarkable degree of sophistication. Most of these approaches, however, typically exploited handcrafted models and knowledge provided by human experts.

More recently, when Machine Learning reached a sufficient maturity to cope with complex data, and, particularly, after the wide spread of Data Mining techniques, temporal and/or spatial data are becoming a primary source for automatic model learning [5]. More specifically, sequences from Molecular Biology (DNA and proteins analysis [6]), computer logs (intrusion detection [7]), Geographical Information Systems (GIS) [8], and images [9] are by now preferred targets for machine learning and knowledge retrieval and discovery.

In AI two broad classes of approaches have been used to represent models of 1-dimensional (1D) or 2-dimensional (2D) data: formal languages (grammars, automata) and logical languages (subsets of predicate logic). In the formal language class we are interested in Hidden Markov Models (HMM), which have proved to be particularly well suited to >>>