Seminar on Computer Science and Applied Mathematics

 

PROGRAM


Matematički Institut
Matematički fakultet
Fakultet organizacionih nauka
JUPIM
IEEE Computer Chapter, Srbija

SEMINAR ZA RAČUNARSTVO I PRIMENJENU MATEMATIKU

MI SANU, Knez Mihailova 36, sala 301f

Upravni odbor Matematickog instituta SANU je na nedavnoj sednici doneo odluku da se dosadasnji Seminar za primenjenu matematiku, sada nazove Seminar za racunarstvo i primenjenu matematiku, a u cilju potenciranja znacaja racunarstva kao jedne od oblasti delatnosti Instituta. Istovremeno, Upravni odbor doneo je odluku o osnivanju Odeljenja za racunarstvo i primenjenu matematiku i vezao rad novog odeljenja za rad Seminara za racunarstvo i primenjenu matematiku.

PLAN RADA SEMINARA ZA AVGUST 2012. GODINE

Utorak, 14.08.2012. u 14:15, Sala 301f, MI SANU:

Predavac: Snezana Minic
Saradnici: Abraham Punnen, Daniel Karapetyan, and Krishna T. Malladi

OPTIMIZACIONI PROBLEM SPUSTANJA SLIKA SA SATELITA NA ZEMLJU: HEURISTIKA BAZIRANA NA OKOLINI 'LANCA IZBACIVANJA' ('EJECTION CHAIN')

Sadrzaj: Sinteticki radar (SAR) omogucava satelitima da efikasno generisu kvalitene slike zemljine povrsine. To generise znacajan komunikacioni problem spustanja slika sa satelita na stanice na zemlji tako da bi se pravilno iskoristele pune mogucnosti ovih sistema, potrebne su efikasne optimizacione metode.

Predavanje predstavlja komunikacioni problem spustanja slika za kanadski SAR satelite, RADARSAT-2. Broj ogranicenja kojima je definisan taj problem iz prakse je znacajan, sto utice na to da je i nalazenje dopustivog resenja kompikovan problem. Heuristika koju smo korsitili za resavanje problema, omogucava nam da generisemo efikasne redoslede spustanja slika i da zadovoljimo sva ogranicenja. Eksperimentalni rezultati na prakticnim problemima pokazuju znacajno poboljsanje u odnosu na prethodno koriscene algoritme.

Speaker: Snezana Minic
Collaborative work done with Abraham Punnen, Daniel Karapetyan, and Krishna T. Malladi

AN EJECTION-CHAIN HEURISTIC FOR THE SATELLITE DOWNLINK SCHEDULING PROBLEM: A CASE STUDY WITH RADARSAT-2

Abstract: The synthetic aperture radar (SAR) technology enables satellites to efficiently acquire high quality images of the Earth surface. To fully utilize the capabilities of these systems it is necessary to have efficient satellite to ground communciations.

This talk addresses the downlink scheduling problem for Canada's Earth observing SAR satellite, RADARSAT-2. Being an applied problem, downlink scheduling is characterized with a number of constraints that make it difficult not only to optimize the schedule but even to produce a feasible solution. We propose a schedule generation procedure that lets us nicely incorporate all the constraints and then effectively optimize the schedule. Our computational experiments conducted on the real data show that the proposed algorithm is a significant improvement over the scheduling procedure previously used.

Ponedeljak, 27.08.2012. u 15h, Sala 301f, MI SANU:

Zoran Obradovic, Director, Center for Data Analytics and Biomedical Informatics, Temple University, Philadelphia, USA
SPATIO-TEMPORAL REGRESSION FOR VARYING QUALITY DATA IN THE PRESENCE OF A LARGE FRACTION OF MISSING VALUES

Abstract: We will first describe our method for spatio-temporal regression from data of varying quality based on Gaussian Conditional Random Fields. When applied to a challenging real-life data of varying quality, this method successfully exploited spatio-temporal properties of observations and outperformed existing alternatives. Next, we will discuss our method for regression from spatio-temporal data with a large fraction of missing values. Our method outperformed alternatives when imputing up to 80% of nonrandom missing values inspatio-temporal multivariate data and accuracy of spatio-temporal regression on such imputed data was much better than when imputing data by nine alternatives. Finally, we will show how to completely avoid the data imputation step when learning form partial observations by taking into account the uncertainty of each instance due to the missing values. Ourrecent method (published last month at AAAI 2012 conference) was shown to outperform the alternatives when there is a large fraction of missing values in data.

Speaker information: Zoran Obradovic is professor of Computer and Information Sciences and the director of the Center for Data Analytics and Biomedical Informatics at Temple University in Philadelphia. His data analytics work is published in more than 260 articles and is cited more than 8,700 times (H-index 40 and I10-index 87).Obradovic is the executive editor at the journal on Statistical Analysis and Data Mining, which is the official publication of the American Statistical Association (ASA) and is currently an editorial board member at eleven journals.Dr. Obradovic is general co-chair for 2013 and 2014 SIAM International Conference on Data Mining and was the program and/or track chair at many data mining and biomedical informatics conference.

RUKOVODIOCI SEMINARA

Vera Kovačević-Vujčić
Milan Dražić