Traffic forecasting is an important research area in Intelligent Transportation Systems that is focused on anticipating traffic in order to mitigate congestion. In this work we propose a deep neural network that simultaneously extracts the spatial features of traffic, using graph convolution, and its temporal features by means of Long Short Term Memory (LSTM) cells to make both short-term and long-term predictions. The model is trained and tested using sparse trajectory (GPS) data coming from the ride-hailing service of DiDi in the cities of Xi'an and Chengdu in China. Besides, presenting the deep neural network, we also propose a data-reduction technique based on temporal correlation to select the most relevant road links to be used as input. Combining the suggested approaches, our model obtains better results compared to high-performance algorithms for traffic forecasting, such as LSTM or the algorithms presented in the TRANSFOR19 forecasting competition. The model is capable of maintaining its performance over different time-horizons from 5 minutes to up to 4 hours with multi-step predictions.
This paper presents a method of optimizing the elements of a hierarchy of Fuzzy Rule-Based Systems (FRBSs). It is a hybridization of a Genetic Algorithm (GA) and the Cross Entropy (CE) method, named GACE. It is used to predict congestion in a 9-km-long stretch of the I5 freeway in California, with time horizons of 5, 15, and 30 minutes. A comparative study of different levels of hybridization in GACE is made. These range from a pure GA to a pure CE, passing through different weights for each of the combined techniques. The results prove that GACE is more accurate than GA or CE alone for predicting short-term traffic congestion.
A real-world newspaper distribution problem with recycling policy is tackled in this work. In order to meet all the complex restrictions contained in such a problem, it has been modeled as a rich vehicle routing problem, which can be more specifically considered as an asymmetric and clustered vehicle routing problem with simultaneous pickup and deliveries, variable costs and forbidden paths (AC-VRP-SPDVCFP). This is the first study of such a problem in the literature. For this reason, a benchmark composed by 15 instances has been also proposed. In the design of this benchmark, real geographical positions have been used, located in the province of Bizkaia, Spain. For the proper treatment of this AC-VRP-SPDVCFP, a discrete firefly algorithm (DFA) has been developed. This application is the first application of the firefly algorithm to any rich vehicle routing problem. To prove that the proposed DFA is a promising technique, its performance has been compared with two other well-known techniques: an evolutionary algorithm and an evolutionary simulated annealing. Our results have shown that the DFA has outperformed these two classic meta-heuristics.
Despite the broad range of Machine Learning (ML) algorithms, there are no clear baselines to find the best method and its configuration given a Short-Term Traffic Forecasting (STTF) problem. In ML, this is known as the Model Selection Problem (MSP). Although Automatic Algorithm Selection (AAS) has proved success dealing with MSP in other areas, it has hardly been explored in STTF. This paper deepens into the benefits of AAS in this field. To this end, we have used Auto-WEKA, a well-known AAS method, and compared it to the general approach (which consists of selecting the best of a set of algorithms) over a multi-class imbalanced classification STTF problem. Experimental results show AAS as a promising methodology in this area and allow important conclusions to be drawn on how to improve the performance of ASS methods when dealing with STTF.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.