published journal article
The research presented here is a continuation of the work published in a previous issue of this journal. The overall objective was to relate travel patterns and urban structure using continuous spatial distributions and urban-economic concepts of residential location choice. In the present paper, model hypotheses are tested using data from a transportation planning study in Washington.
presentation
conference paper |
Jan 2023
conference paper
With the rapid evolution of the Internet of Things, many real-world applications utilize heterogeneously connected sensors to capture time-series information. Edge-based machine learning (ML) methodologies are often employed to analyze locally collected data. However, a fundamental issue across data-driven ML approaches is distribution shift. It occurs when a model is deployed on a data distribution different from what it was trained on, and can substantially degrade model performance. Additionally, increasingly sophisticated deep neural networks (DNNs) have been proposed to capture spatial and temporal dependencies in multi-sensor time series data, requiring intensive computational resources beyond the capacity of today’s edge devices. While brain-inspired hyperdimensional computing (HDC) has been introduced as a lightweight solution for edge-based learning, existing HDCs are also vulnerable to the distribution shift challenge. In this paper, we propose DOMINO, a novel HDC learning framework addressing the distribution shift problem in noisy multi-sensor time-series data. DOMINO leverages efficient and parallel matrix operations on high-dimensional space to dynamically identify and filter out domain-variant dimensions. Our evaluation on a wide range of multi-sensor time series classification tasks shows that DOMINO achieves on average 2.04% higher accuracy than state-of-the-art (SOTA) DNN-based domain generalization techniques, and delivers 16.34times faster training and 2.89times faster inference. More importantly, DOMINO exhibits notably better performance when learning from partially labeled data and highly imbalanced data, and provides 10.93times higher robustness against hardware noises than SOTA DNNs.
published journal article
In this paper, the authors mathematically model the multi-hop Peer-to-Peer (P2P) ride-matching problem as a binary program. The authors formulate this problem as a many-to-many problem in which a rider can travel by transferring between multiple drivers, and a driver can carry multiple riders. The authors propose a pre-processing procedure to reduce the size of the problem, and devise a decomposition algorithm to solve the original ride-matching problem to optimality by means of solving multiple smaller problems. The authors conduct extensive numerical experiments to demonstrate the computational efficiency of the proposed algorithm and show its practical applicability to reasonably-sized dynamic ride-matching contexts. Finally, in the interest of even lower solution times, the authors propose heuristic solution methods, and investigate the trade-offs between solution time and accuracy.
research report
published journal article
published journal article
published journal article
Background: Although numerous epidemiologic studies now use models of intraurban exposure, there has been little systematic evaluation of the performance of different models. Objectives: In this present article we proposed a modeling framework for assessing exposure model performance and the role of spatial autocorrelation in the estimation of health effects. Methods: We obtained data from an exposure measurement substudy of subjects from the Southern California Children’s Health Study. We examined how the addition of spatial correlations to a previously described unified exposure and health outcome modeling framework affects estimates of exposure-response relationships using the substudy data. The methods proposed build upon the previous work, which developed measurement-error techniques to estimate long-term nitrogen dioxide exposure and its effect on lung function in children. In this present article, we further develop these methods by introducing between- and within-community spatial autocorrelation error terms to evaluate effects of air pollution on forced vital capacity. The analytical methods developed are set in a Bayesian framework where multistage models are fitted jointly, properly incorporating parameter estimation uncertainty at all levels of the modeling process. Results: Results suggest that the inclusion of residual spatial error terms improves the prediction of adverse health effects. These findings also demonstrate how residual spatial error may be used as a diagnostic for comparing exposure model performance.
published journal article
Planning decisions involving urban freeways have drastically affected American cities by reconfiguring urban form, supplanting neighborhoods, displacing tens of thousands of people, and costing billions of dollars. New laws governing the planning and construction of new freeways were passed that required projects to factor in maximum sensitivity to environmental effects, concern for relocating displaced residents, and active citizen participation. This article reviews the evolution of freeway design in response to significant social changes in the United States, and examines changes in the rgulatory environment of freeway construction. The article describes three famous cases of urban freeway controversies in Los Angeles, Memphis, and Syracuse which have distinct histories and outcomes. From these cases it can be concluded that the past mode of highway planning was too narrow and now multimodal transportation planning involving the equal partnership of trasportation planners, land use planners, urban planners, and urban designers should be utilized.
published journal article
This paper uses a before-after, experimental-control group method to evaluate the impacts of the newly opened Expo light rail transit line in Los Angeles on personal vehicle greenhouse gas (GHG) emissions. We applied the California Air Resources Board’s EMFAC 2011 emission model to estimate the amount of daily average CO2 emissions from personal vehicle travel for 160 households across two waves, before and after the light rail opened. The 160 households were part of an experimental-control group research design. Approximately half of the households live within a half-mile of new Expo light rail stations (the experimental group) and the balance of the sampled households live beyond a half-mile from Expo light rail stations (the control group). Households tracked odometer mileage for all household vehicles for seven days in two sample waves, before the Expo Line opened (fall, 2011) and after the Expo Line opened (fall, 2012). Our analysis indicates that opening the Expo Line had a statistically significant impact on average daily CO2 emissions from motor vehicles. We found that the CO2 emission of households who reside within a half-mile of an Expo Line station was 27.17 percent smaller than those living more than a half-mile from a station after the opening of the light rail, while no significant difference exists before the opening. A difference-in-difference model suggests that the opening of the Expo Line is associated with 3,145 g less of household vehicle CO2 emissions per day as a treatment effect. A sensitivity analysis indicates that the emission reduction effect is also present when the experimental group of households is redefined to be those living within a kilometer from the new light rail stations.