[Solomonov Seminar] 208. Solomonov seminar

[Marko Grobelnik]

V torek 19. maja bo ob 13h v Veliki predavalnici IJS
(prvo nadstropje glavne zgradbe), 208. Solomonov seminar.
Posnetki preteklih seminarjev so na http://videolectures.net/solomon/

Predavala bo Biljana Mileva Boskoska iz Univerze Cirila in Metoda 
v Skopju in IJS-MPS o napovedovanju onesnazenja zraka.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Biljana Mileva Boskoska:
  Comparison of prediction models for air pollution using SVM and RBF neural networks
  
Science deals with modeling of ecological systems in order to explore or improve the 
environmental conditions in which we live. Air is one segment in the environment that 
is very important for the health of people and it is widely used by scientists for prediction 
of its pollution. 

The problem of air pollution is a worldwide problem that exists for centuries. Recently, 
a lot of studies have been done on examining the impacts of pollution of ozone and nitrogen 
oxides onto the health of people and the environment. The main observed problems are in 
the process of precise and on-time monitoring. Our goal is to define tools for prediction 
of air pollution based on historical data. We used two techniques: support vector machines 
and radial basis function neural networks. Using the data from one monitoring station, 
we evaluate the efficiency and the accuracy of both approaches in the process of predicting 
the air pollution in the next 24 to 240 hours. A detailed overview of the parameter selection 
process is given for each of the approaches.