By Andrea Ferrara, Paolo Liberatore, Marco Schaerf (auth.), Stefania Bandini, Sara Manzoni (eds.)
This quantity collects the papers chosen for presentation on the IX Congress of the Italian organization for Arti?cial Intelligence (AI*IA), held in Milan on the college of Milano–Bicocca (September 21–23, 2005). at the one hand this congress maintains the culture of AI*IA in organizing its biannual s- enti?c assembly from 1989; nonetheless, this variation is a landmark within the involvement of the overseas neighborhood of arti?cial intelligence (AI), without delay related to a wide variety of specialists from a number of international locations within the P- gramCommittee. Moreover,the ordinary nature of scienti?c researchin arti?cial intelligence (which is intrinsically foreign) and several other consolidated int- nationwide collaborations in initiatives and mobility courses allowed the gathering and choice of papers from many di?erent international locations, everywhere in the international, enlarging the visibility of the Italian contribution inside of this examine ?eld. Arti?cial intelligence is at the present time a growing to be advanced set of conceptual, theor- ical, methodological, and technological frameworks, o?ering cutting edge com- tational ideas within the layout and improvement of computer-based structures. inside of this attitude, researchers operating during this sector needs to take on a wide diversity of information approximately equipment, effects, and recommendations coming from di?erent classical parts of this self-discipline. The congress was once designed as a discussion board permitting researchers to offer and talk about really expert effects as normal contributions to AI growth.
Read or Download AI*IA 2005: Advances in Artificial Intelligence: 9th Congress of the Italian Association for Artificial Intelligence, Milan, Italy, September 21-32, 2005. Proceedings PDF
Best italian books
Extra resources for AI*IA 2005: Advances in Artificial Intelligence: 9th Congress of the Italian Association for Artificial Intelligence, Milan, Italy, September 21-32, 2005. Proceedings
The variable that produces the largest increment in the number of satisﬁed clauses is ﬂipped (no random walk is performed). WalkSAT has usually a far better performance than GSAT. Nevertheless, both algorithms lack a global strategy that could guide them during the exploration of the search space. Algorithms equipped with such a strategy are commonly called metaheuristics . In order to extend the diversity of the techniques compared, we applied also an ILS designed to attack SAT and MAXSAT problems [14,15].
We can observe the typical behavior of instances interpolating between regular and random instances. 5 0 0 50 100 150 200 250 Number of clauses from RandomSAT 300 350 0 100 600 700 800 variables, 2400 clauses 10 10 9 9 8 8 7 7 Proximity ratio Proximity ratio 500 variables, 1500 clauses 200 300 400 500 Number of clauses from RandomSAT 6 5 4 6 5 4 3 3 2 2 1 1 0 0 0 200 400 600 800 1000 1200 Number of clauses from RandomSAT 1400 1600 0 500 1000 1500 2000 Number of clauses from RandomSAT 2500 Fig. 8.
Points represent median iterations over 1000 runs. Log-scale on the y-axis has been used when necessary. respectively. , the instances with strong lattice/small-world topologies3. In some plots, we also observe that the instances corresponding to the maximal proximity ratio are the hardest on average4. Nevertheless, this behavior is not regular nor clear and the statistical correlation between search cost and proximity ratio is quite low. The peculiar behavior observed is a clear signal that different factors other than small-world topology affect algorithm behavior.
AI*IA 2005: Advances in Artificial Intelligence: 9th Congress of the Italian Association for Artificial Intelligence, Milan, Italy, September 21-32, 2005. Proceedings by Andrea Ferrara, Paolo Liberatore, Marco Schaerf (auth.), Stefania Bandini, Sara Manzoni (eds.)