Grzegorz Swirszcz

Traffic Velocity Prediction Using GPS Data: IEEE ICDM Contest Task 3 Report

This report summarizes the methodologies and techniques we developed and applied for tackling task 3 of the IEEE ICDM Contest on predicting traffic velocity based on GPS data. The major components of our solution include 1) A pre-processing procedure to map GPS data to the network, 2) A K-nearest neighbor approach for identifying the most similar training hours for every test hour, and 3) A heuristic evaluation framework for optimizing parameters and avoiding over-fitting. Our solution finished Second in the final evaluation.

Ensemble-Based Method for Task 2: Predicting Traffic Jam

In this paper, we describe our solution for ICDM 2010 Contest Task 2 (Jams), where the task is to predict future where the next traffic jams will occur in morning rush hour, given data gathered during the initial phase of this peak period. Our solution, which is based on an ensemble approach, finished Second in the final evaluation.

Inventory allocation and transportation scheduling for logistics of network-centric military operations

This paper describes a prototype inventory-placement and transportation-scheduling solution developed in support of the emerging military doctrine of Network-Centric Operations (NCO). NCO refers to an unprecedented ability to share information among cooperating forces, enabled by modern communications and computing technology. The objective of the Network-Centric concept is to collect, disseminate, and react to real-time information in order to improve the performance of the U.S. Army as a fighting force. One problem that arises in the logistics domain involves the maintenance of combat vehicles. We seek to determine the improvement, if any, made possible by exploiting accurate information on the status of available repair parts inventory, the current locations of mobile supply points, and the demand for parts. We describe logistics algorithms for maximizing the operational availability of combat vehicles by producing, flexible, optimized inventory and delivery plans that decrease replenishment times and prioritize parts allocations and repairs. Our algorithms are designed to leverage real-time information available from modern communications and inventory tracking technology by employing state-of-the-art mathematical optimization models. Our simulations indicate that Network-Centric Logistics (NCL) can significantly improve combat vehicle availability in comparison with current practice.

Distance Geometry on the Sphere

The Distance Geometry Problem asks whether a given weighted graph has a realization in a target Euclidean space \(\mathbb {R}^K\) which ensures that the Euclidean distance between two realized vertices incident to a same edge is equal to the given edge weight. In this paper we look at the setting where the target space is the surface of the sphere \(\mathbb {S}^{K-1}\). We show that the Distance Geometry Problem is almost the same in this setting, as long as the distances are Euclidean. We then generalize a theorem of Godel about the case where the distances are spherical geodesics, and discuss a method for realizing cliques geodesically on a K-dimensional sphere.