Search

The causal inference is the process of determining the causality and correlation between different variables in a system. This paper mainly focuses on the introduction of theories of causal inference analysis and application of causal inference on student performance data. The adjustment formula, backdoor criterion and front-door criterion are three different methods to analyze different causal diagrams which can be used to quantify and analyze the causal effect of students' choices on performance and school support. Taking advantage of logistic regression model, counterfactual inference can utilize students' experience to predict the possibility of result under different choice conditions. The Natural Direct Effect in the path-specific effect is used to study the direct effect of sensitive attribute gender on the acquisition of school support, determining the fairness of decision-making of the school support

Bus rapid transit (BRT) has been proved to be an effective tool to improve mass transit services. However, BRT's adaptive operations like management and scheduling under different scenarios are too complicated to implement using traditional methods. The ACP approach, which is based on holism and complex system theory and consists of artificial systems (A), computational experiments (C) and parallel execution (P), offers an efficient new method to cope with these complex systems, including BRT. In this paper, the parallel transportation management and control system for BRT (PTMS-BRT) is presented, which is designed and implemented using the ACP approach. PTMS-BRT integrates such functions as BRT's monitoring, warning, forecasting, incident management, and real-time scheduling, to provide its operations smoother, safer, more efficient, and reliable. It has been piloted successfully in Guangzhou BRT to demonstrate it as another successful example of parallel transportation systems.

Artificial intelligence, led by the method of deep learning, has generated enormous interest in both professional circle and general public in the last two years thanks to Deepmind’s AlphaGo’s stunning mastery of Go, the most sophisticated board game. While most interest since then has been shown in exploring the applications of AlphaGo’s algorithms in machine learning, it is the potential impact of its learning strategy on human learning that captures our attention. Can AlphaGo’s success, aside from taking advantage of superior computing power, lead to more effective learning for humans? Does AlphaGo’s learning lend support to any of the learning theories? Or does the training data reveal any notable pattern or trajectory that may suggest new perspectives on human cognition? In this work-in-progress paper, we try to make connection between human and machine learning using the technical details revealed by the Deepmind team, and examine what insights can be gained from AlphaGo’s training on human cognitive development and more specifically, engineering education.

This paper discusses the trend modeling for traffic time series. First, we recount two types of definitions for a long-term trend that appeared in previous studies and illustrate their intrinsic differences. We show that, by assuming an implicit temporal connection among the time series observed at different days/locations, the PCA trend brings several advantages to traffic time series analysis. We also describe and define the so-called short-term trend that cannot be characterized by existing definitions. Second, we sequentially review the role that trend modeling plays in four major problems in traffic time series analysis: abnormal data detection, data compression, missing data imputation, and traffic prediction. The relations between these problems are revealed, and the benefit of detrending is explained. For the first three problems, we summarize our findings in the last ten years and try to provide an integrated framework for future study. For traffic prediction problem, we present a new explanation on why prediction accuracy can be improved at data points representing the short-term trends if the traffic information from multiple sensors can be appropriately used. This finding indicates that the trend modeling is not only a technique to specify the temporal pattern but is also related to the spatial relation of traffic time series.

In this paper, we focus on a special problem in transportation studies that concerns the so called “Big Data” challenge, which is: how to build concise yet accurate traffic flow prediction models based on the massive data collected by different sensors ? The size of the data, the hidden causal dependence and the complexity of traffic time series are some of the obstacles that affect making reliable forecast at a reasonable cost, both time-wise and computation-wise. To better prepare the data for traffic modeling, we introduce a multiple-step strategy to process the raw “Big Data” into compact time series that are better suited for regression and causality analysis. First, we use the Granger causality to define and determine the potential dependence among data, and produce a much condensed set of times series who are also highly dependent. Next, we deploy a decomposition algorithm to separate daily-similar trend and nonstationary bursts components from the traffic flow time series yielded by the Granger test. The decomposition results are then treated by two rounds of Lasso regression: the standard Lasso method is first used to quickly filter out most of the irrelevant data, followed by a robust Lasso method to further remove the disturbance caused by bursts components and recover the strongest dependence among the remaining data. Test results show that the proposed method significantly reduces the costs of building prediction models. Moreover, the obtained causal dependence graph reveals the relationship between the structure of road networks and the correlations among traffic time series. All these findings are useful for building better traffic flow prediction models.

This paper investigates the productivity and collaborations patterns in the publications of IEEE Transactions on Intelligent Transportation Systems (T-ITS) between 2010 and 2013. Our findings show that industry is playing a more important role in research cooperation. Such a trend is particularly beneficial for expediting the commercialization of scientific and technological advances. In addition, growing international collaborations have become a norm as researchers in the U.S., China, and Europe are increasingly involved in joint research endeavors.

Preserving important troubleshooting experience gained in industrial practice, often referred to as “tribal knowledge”, and making it readily available to be re-applied in similar situations represent a serious challenge for many companies in heavy industry. The solution is very often tied to building mental models to capture and express key latent cognitive variables. In this paper, we first attempt to identify the appropriate knowledge-based system diagnosis paradigms. A conceptual map-based framework is then proposed as a potential solution for retaining tribal knowledge. The validity and feasibility of the approach are demonstrated through a training program developed under the support of National Science Foundation (NSF) for improving engineering and technology students' troubleshooting skills.

Cardiovascular diseases are the leading cause of death in the United States. Advances in wireless technology have made possible the remote monitoring of a patient's heart sensors as part of a body area network. Some of these networks use a Bluetooth Low Energy (BLE) component to transmit the signal between the bio sensors and a smart phone. This study investigates the impact that stray wireless transmissions from residential microwave ovens have on the BLE component of the body area network. The results of this study may lead to improvements and widespread use of body area network medical systems, which may lead to better monitoring, more data, and fewer fatalities due to misdiagnosed heart conditions.

The application of supervised learning to train an intelligent vehicle with a neuro-fuzzy controller to mimic the driving behavior of a human driver is discussed. An initial fuzzy control system for vehicle driving was set up on the basis of general human driving experiences, and its control rules were modified to fit the driving behavior of an individual driver. This provides an effective mechanism to construct driving control systems with personality for automated intelligent vehicles.

This article describes the architecture of a web-based audio/video educational system for real-time laboratory experiments through Internet, currently developed at the WAVES Lab of the University of Arizona, Tucson, Arizona. The goal of the WAVES project is to develop an open and integrated platform for online laboratory experimentation, computer simulation, and course instruction using the Internet for research and teacfiing in modeling, design, analysis, and control of dynamical systems. The project is unique in that we can conduct hardware lab experiments via the Internet and provide researchers and students at remote sites with real-time data feedback and delayed (few minutes) audio/video feedback of experimented results. Although many Internet-based tools have been developed for instruction, to the best of our knowledge, none includes “hand-on” experiments with actual hardware. Since Internet-based hardware systems can be shared and scheduled for use by a large number of researchers and students in various locations, the WAVES platform is a cost-effective way to address the issue of “hands on” experience for students and researchers. In our current implementation, a three layer architecture is used, i.e., the clients, the dispatcher, and the coordinators. First, the clients issue tasks and receive feedback for conducting a specific lab experiment; then the dispatcher deals with the control and communication between the clients and coordinators; finally the coordinators conduct the actual experiment through hardware setups, simulation models, and database systems.

Accurate assessment and effective control of urban passenger transportation hub operations are an area that traditional methods have yet to address satisfactorily. This article presents a framework for parallel management and control based on the artificial societies, computational experiments, and parallel execution (ACP) paradigm. The authors first introduce the components, internal interactions, and support mechanism of artificial transportation hubs. They then describe an agent-based model for the hub including the computational experiments and responsibilities of parallel execution in active hub management and control. Thy also present an implementation of the framework through field tests. The framework can improve operational efficiency and service quality of urban passenger transportation hubs.

This paper proposes a novel approach to constructing fuzzy logic control system using the combination of wavelet network and feedforward BP networks. Numerous works have been reported on the integration, of fuzzy logic systems and neural networks; however most of these hybridizations have resulted in the loss of the original knowledge structure embedded in its network implementation. Due to the inherent modularity, the decision-making process of a fuzzy logic control system can be readily partitioned into three functionally independent modules, namely pattern recognition, fuzzy reasoning and control synthesis, and therefore individual subnets can be built to implement these modules as subnets for the overall feedforward networks. Both wavelet network and BP feedforward network are adopted for the construction of the subnets so that the local receptive field-type wavelet networks with orthogonal least square learning algorithm are capable to capture the fine details of the underlying dynamics, and the overall feedforward structure with the BP algorithm is responsible for the global optimization of the networks parameters. Simulation also shows its superiority.

The proposed system architecture presents a mobile-agent based three-tier distributed control architectural framework in order to provide QoS-driven, scalable system configuration, flexible integration of heterogeneous control systems and robust hierarchical distributed control system for home automation and intelligent control. In this architecture, CORBA and mobile agent technologies work together as a flexible middle layer to provide the distribution of service control and location-aware adaptability and achieve large accessibility and interoperability for distributed control. To satisfy real-time requirements of home automation, the priority queue scheduling, real time event dispatching mechanisms and HAN architecture are also presented.

This paper describes the architecture of a Web-based audio/video educational system for real-time laboratory experiments through the Internet. The goal of the WAVES (Web-based Audio Video Educational Systems) project is to develop an open and integrated platform for online laboratory experimentation, computer simulation, and course instruction using the Internet for research and teaching in modeling, design, analysis, and control of dynamical systems. The project is unique in that students can conduct hardware lab experiments via the Internet and receive real-time data feedback and audio/video feedback of experimented results. Although many similar Internet-based tools have been developed throughout the last two years, very few of them encompass "hands-on" experiments with actual laboratory hardware. Since Internet-based hardware systems can be shared and scheduled for use by a large number of researchers and students in various locations, the WAVES platform is a cost-effective way to address the issue of "hands-on" experience for students. In our current implementation, a three layer architecture is used, i.e., the clients, the dispatcher, and the coordinators. First, the clients will issue tasks and receive feedback for conducting a specific lab experiment; then the dispatcher deals with the control and communication between the clients and coordinators; finally the coordinators conduct the actual experiment through hardware setups, simulation models, and database systems.

This paper outlines a computational theory of linguistic dynamic systems for computing with words by fusing procedures and concepts from several different areas: Kosko's (1992) geometric interpretation of fuzzy sets, Hsu's (1987) cell-to-cell mappings in nonlinear analysis, equi-distribution lattices in number theory, and dynamic programming in optimal control theory. The proposed framework enables us to conduct a global dynamic analysis, system design and synthesis for linguistic dynamic systems that use words or linguistic terms in computation, based on concepts and methods well developed for conventional dynamic systems. This theory has significant potential for modeling and analysis of systems where model, goal, control and feedback are primarily specified in words or linguistic terms.

This paper outlines the architecture and state of art technology of utilizing computer communication networks and field programmable logic devices to implement intelligent neuro-fuzzy control systems for smart consumer electronics.