Low-Rank Optimal Transport for Robust Domain Adaptation

When encountering the distribution shift between the source(training) and target(test) domains, domain adaptation attempts to adjust the classifiers to be capable of dealing with different domains. Previous domain adaptation research has achieved a lot of success both in theory and practice under the assumption that all the examples in the source domain are welllabeled and of high quality. However, the methods consistently lose robustness in noisy settings where data from the source domain have corrupted labels or features which is common in reality. Therefore, robust domain adaptation has been introduced to deal with such problems. In this paper, we attempt to solve two interrelated problems with robust domain adaptation:distribution shift across domains and sample noises of the source domain. To disentangle these challenges, an optimal transport approach with low-rank constraints is applied to guide the domain adaptation model training process to avoid noisy information influence. For the domain shift problem, the optimal transport mechanism can learn the joint data representations between the source and target domains using a measurement of discrepancy and preserve the discriminative information. The rank constraint on the transport matrix can help recover the corrupted subspace structures and eliminate the noise to some extent when dealing with corrupted source data. The solution to this relaxed and regularized optimal transport framework is a convex optimization problem that can be solved using the Augmented Lagrange Multiplier method, whose convergence can be mathematically proved. The effectiveness of the proposed method is evaluated through extensive experiments on both synthetic and real-world datasets.

A Heterogeneous Information Fusion Method for Maritime Radar and AIS Based on D-S Evidence Theory

Maritime radar and automatic identification systems (AIS), which are essential auxiliary equipment for navigation safety in the shipping industry, have played significant roles in maritime safety supervision. However, in practical applications, the information obtained by a single device is limited, and it is necessary to integrate the information of maritime radar and AIS messages to achieve better recognition effects. In this study, the D-S evidence theory is used to fusion the two kinds of heterogeneous information: maritime radar images and AIS messages. Firstly, the radar image and AIS message are processed to get the targets of interest in the same coordinate system. Then, the coordinate position and heading of targets are chosen as the indicators for judging target similarity. Finally, a piece of D-S evidence theory based on the information fusion method is proposed to match the radar target and the AIS target of the same ship. Particularly, the effectiveness of the proposed method has been validated and evaluated through several experiments, which proves that such a method is practical in maritime safety supervision.

Degradation of Silicone Rubbers in Fenton's Reagents

Gaskets are applied in PEMFCs（proton exchange membrane fuel cells） to keep reactant gases and liquid within their respective regions, which are of great significance for the both sealing and electrochemical performance of fuel cells during the long-term operation. In this study, the degradation of silicone rubbers, often selected as seals in PEMFCs, in Fenton＇s reagents with different H_2O_2 concentrations was investigated. The changes in chemical properties, mechanical behavior and surface morphology of the samples were studied before and after exposure to the test environment over time. It is found that increasing H_2O_2 concentration will degrade the rubbers more severely. The experimental results elucidate the degradation mechanism of silicone rubbers in Fenton＇s reagents and the influence of H_2O_2 in the degradation process.

Event-triggered neuroadaptive predefined practical finite-time control for dynamic positioning vessels:A time-based generator approach

This paper discusses the predefined practical finite-time(PPFT)dynamic positioning(DP)control problem for DP vessels subject to internal/external uncertainties.Those heterogeneity uncertainties are handled by a separate-type treatment approach.The finite-time(FT)DP control is fulfilled by a predefined FT function on the basis of a time-based generator(TBG).Under the dynamic surface control together with the TBG design framework,the convergence time and control accuracy of the DP system can be determined by the designer offline.Meanwhile,the virtual derivation and computational burden problems are dissolved by using a first-order filter and virtual parameter learning technique.To reduce mechanical wear,an event-triggering protocol between the control law and the actuator is built to reduce the operating frequency of the actuator.An event-triggered neuroadaptive PPFT control scheme is presented for DP vessels.The stability of the closed-loop DP control systems is validated via the Lyapunov theorem.Approach efficiency is confirmed by numerical examples.

Effects of winter weather on traffic operations and optimization of signalized intersections

Adverse winter weather has always been a cause of traffic congestion and road collisions.To mitigate the negative impacts of winter weather, transportation agencies are under increasing pressure to introduce weather responsive traffic management strategies.Currently, most traffic signal control systems are designed for normal weather conditions and are therefore suboptimal regarding efficiency and safety for controlling traffic during winter snow events due to changes in traffic patterns and driver behaviors. The main objective of this research is to explore how to modify pre-timed traffic signal control parameters under adverse weather conditions to increase traffic efficiency and road safety.This research consists of two main components. First, we examine the impacts of winter weather on three key traffic parameters, i.e., saturation flow rate, start-up lost time, and free flow speed. Secondly, we investigate the potential benefits of implementing weatherspecific signal control plans for uncoordinated intersections as well as coordinated corridors. Two case studies are conducted, each with varying levels of traffic demand and winter event severity, to compare the performance of different signal plans. Evaluation results from both Synchro and VISSIM show that implementing such signal plans is most beneficial for intersection with a medium level of traffic demand. It is also found that the benefit of implementing weather-responsive plans was more compelling at a coordinatedcorridor level than at an uncoordinated-intersection level.

Using naturalistic driving data to identify driving style based on longitudinal driving operation conditions

Purpose–An individual’s driving style significantly affects overall traffic safety.However,driving style is difficult to identify due to temporal and spatial differences and scene heterogeneity of driving behavior data.As such,the study of real-time driving-style identification methods is of great significance for formulating personalized driving strategies,improving traffic safety and reducing fuel consumption.This study aims to establish a driving style recognition framework based on longitudinal driving operation conditions(DOCs)using a machine learning model and natural driving data collected by a vehicle equipped with an advanced driving assistance system(ADAS).Design/methodology/approach–Specifically,a driving style recognition framework based on longitudinal DOCs was established.To train the model,a real-world driving experiment was conducted.First,the driving styles of 44 drivers were preliminarily identified through natural driving data and video data;drivers were categorized through a subjective evaluation as conservative,moderate or aggressive.Then,based on the ADAS driving data,a criterion for extracting longitudinal DOCs was developed.Third,taking the ADAS data from 47 Kms of the two test expressways as the research object,six DOCs were calibrated and the characteristic data sets of the different DOCs were extracted and constructed.Finally,four machine learning classification(MLC)models were used to classify and predict driving style based on the natural driving data.Findings–The results showed that six longitudinal DOCs were calibrated according to the proposed calibration criterion.Cautious drivers undertook the largest proportion of the free cruise condition(FCC),while aggressive drivers primarily undertook the FCC,following steady condition and relative approximation condition.Compared with cautious and moderate drivers,aggressive drivers adopted a smaller time headway(THW)and distance headway(DHW).THW,time-to-collision(TTC)and DHW showed highly significant differences in driving style identification,while longitudinal acceleration(LA)showed no significant difference in driving style identification.Speed and TTC showed no significant difference between moderate and aggressive drivers.In consideration of the cross-validation results and model prediction results,the overall hierarchical prediction performance ranking of the four studied machine learning models under the current sample data set was extreme gradient boosting>multi-layer perceptron>logistic regression>support vector machine.Originality/value–The contribution of this research is to propose a criterion and solution for using longitudinal driving behavior data to label longitudinal DOCs and rapidly identify driving styles based on those DOCs and MLC models.This study provides a reference for real-time online driving style identification in vehicles equipped with onboard data acquisition equipment,such as ADAS.