MACHINE LEARNING SOLUTIONS FOR TRANSPORTATION NETWORKS

This thesis brings a collection of novel models and
methods that result from a new look at practical
problems in transportation through the prism of newly
available sensor data.

From this data, we build a model of traffic flow
inspired by macroscopic flow models. Unlike
traditional such models, our model deals with
uncertainty of measurement and unobservability of
certain important quantities and incorporates
on-the-fly observations more easily. Having a
predictive distribution of traffic state enables the
application of powerful decision-making machinery to
the traffic domain.

Secondly, a new method for detecting accidents and
other adverse events is described. Data collected
from highways enables us to bring supervised learning
approaches to incident detection. However, a major
hurdle to performance of supervised learners is the
quality of data which contains systematic biases
varying from site to site. We build a dynamic
Bayesian network framework that learns and rectifies
these biases, leading to improved supervised
detector performance with little need for manually
tagged data. The realignment method applies generally
to virtually all forms of labeled sequential data.

Autorentext

Tomas specializes in machine learning and anomaly detection,especially by means of graphical probability models. He obtainedhis PhD from University of Pittsburgh in 2008, authored paperson inference in graphical models, modeling of large sensor datasets and anomaly detection and served on program committees ofseveral major AI conferences.

Klappentext

This thesis brings a collection of novel models and methods that result from a new look at practical problems in transportation through the prism of newly available sensor data. From this data, we build a model of traffic flow inspired by macroscopic flow models. Unlike traditional such models, our model deals with uncertainty of measurement and unobservability of certain important quantities and incorporates on-the-fly observations more easily. Having a predictive distribution of traffic state enables the application of powerful decision-making machinery to the traffic domain. Secondly, a new method for detecting accidents and other adverse events is described. Data collected from highways enables us to bring supervised learning approaches to incident detection. However, a major hurdle to performance of supervised learners is the quality of data which contains systematic biases varying from site to site. We build a dynamic Bayesian network framework that learns and rectifies these biases, leading to improved supervised detector performance with little need for manually tagged data. The realignment method applies generally to virtually all forms of labeled sequential data.