Ali Haghani

Bio: Ali Haghani is an academic researcher from University of Maryland, College Park. The author has contributed to research in topics: Heuristic (computer science) & Vehicle routing problem. The author has an hindex of 29, co-authored 148 publications receiving 3871 citations. Previous affiliations of Ali Haghani include University of New Hampshire.

A gradient boosting method to improve travel time prediction

Abstract: Tree based ensemble methods have reached a celebrity status in prediction field. By combining simple regression trees with ‘poor’ performance, they usually produce high prediction accuracy. In contrast to other machine learning methods that have been treated as black-boxes, tree based ensemble methods provide interpretable results, while requiring little data preprocessing, are able to handle different types of predictor variables, and can fit complex nonlinear relationship. These properties make the tree based ensemble methods good candidates for solving travel time prediction problems. However, applications of tree-based ensemble algorithms in traffic prediction area are limited. In this paper, we employ a gradient boosting regression tree method (GBM) to analyze and model freeway travel time to improve the prediction accuracy and model interpretability. The gradient boosting tree method strategically combines additional trees by correcting mistakes made by its previous base models, therefore, potentially improves prediction accuracy. Different parameters’ effect on model performance and correlations of input–output variables are discussed in details by using travel time data provided by INRIX along two freeway sections in Maryland. The proposed method is, then, compared with another popular ensemble method and a bench mark model. Study results indicate that the GBM model has its considerable advantages in freeway travel time prediction.

Formulation and solution of a multi-commodity, multi-modal network flow model for disaster relief operations

Abstract: This paper presents a formulation and two solution methods for a very complex logistical problem in disaster relief management. The problem to be addressed is a large-scale multicommodity, multi-modal network flow problem with time windows. Due to the nature of this problem, the size of the optimization model which results from its formulation grows extremely rapidly as the number of modes and/or commodities increase. The formulation of the problem is based on the concept of a time-space network. Two heuristic algorithms are proposed. One is a heuristic which exploits an inherent network structure of the problem with a set of side constraints and the other is an interactive fix-and-run heuristic. The findings of the model implementation are also presented using artificially generated data sets. The performance of the solution methods are examined over a range of small and large problems.

Data Collection of Freeway Travel Time Ground Truth with Bluetooth Sensors

Abstract: Accurate travel time information is essential to the effective management of traffic conditions. Traditionally, floating car data have been used as the primary source of ground truth for measuring the quality of real-time travel time provided by traffic surveillance systems. This paper introduces Bluetooth sensors as a new and effective means of data collection of freeway ground truth travel time. The concept of vehicle identification using Bluetooth signatures for travel time estimation along a section of freeway is explained. Issues related to error analysis, filtering of raw matched data, and accuracy of the resulting ground truth compared with floating car are discussed. Data from loop detectors on several freeway segments are used to approximate and report the average sampling rate of Bluetooth sensors. Results show that the new technology is a promising method for collecting high-quality travel time data that can be used as ground truth for evaluating other sources of travel time and other intellige...

Modeling integrated supply chain logistics in real-time large-scale disaster relief operations

Abstract: The goal of this research is to develop a comprehensive model that describes the integrated logistics operations in response to natural disasters. We propose a mathematical model that controls the flow of several relief commodities from the sources through the supply chain and until they are delivered to the hands of recipients. The structure of the network is in compliance with FEMA's complex logistics structure. The proposed model not only considers details such as vehicle routing and pick up or delivery schedules; but also considers finding the optimal locations for several layers of temporary facilities as well as considering several capacity constraints for each facility and the transportation system. Such an integrated model provides the opportunity for a centralized operation plan that can eliminate delays and assign the limited resources to the best possible use. A set of numerical experiments is designed to test the proposed formulation and evaluate the properties of the optimization problem. The numerical analysis shows the capabilities of the model to handle the large-scale relief operations with adequate details. However, the problem size and difficulty grows rapidly by extending the length of the operations or when the equity among recipients is considered. In these cases, it is suggested to find fast solution algorithms and heuristic methods in future research.

Pattern Recognition and Machine Learning

Abstract: Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

LSTM network: a deep learning approach for short-term traffic forecast

Abstract: Short-term traffic forecast is one of the essential issues in intelligent transportation system. Accurate forecast result enables commuters make appropriate travel modes, travel routes, and departure time, which is meaningful in traffic management. To promote the forecast accuracy, a feasible way is to develop a more effective approach for traffic data analysis. The availability of abundant traffic data and computation power emerge in recent years, which motivates us to improve the accuracy of short-term traffic forecast via deep learning approaches. A novel traffic forecast model based on long short-term memory (LSTM) network is proposed. Different from conventional forecast models, the proposed LSTM network considers temporal-spatial correlation in traffic system via a two-dimensional network which is composed of many memory units. A comparison with other representative forecast models validates that the proposed LSTM network can achieve a better performance.

CatBoost: unbiased boosting with categorical features

Abstract: This paper presents the key algorithmic techniques behind CatBoost, a new gradient boosting toolkit. Their combination leads to CatBoost outperforming other publicly available boosting implementations in terms of quality on a variety of datasets. Two critical algorithmic advances introduced in CatBoost are the implementation of ordered boosting, a permutation-driven alternative to the classic algorithm, and an innovative algorithm for processing categorical features. Both techniques were created to fight a prediction shift caused by a special kind of target leakage present in all currently existing implementations of gradient boosting algorithms. In this paper, we provide a detailed analysis of this problem and demonstrate that proposed algorithms solve it effectively, leading to excellent empirical results.