A. James Clark School of Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/1654
The collections in this community comprise faculty research works, as well as graduate theses and dissertations.
Browse
2 results
Search Results
Item Security Enhancement and Bias Mitigation for Emerging Sensing and Learning Systems(2021) Chen, Mingliang; Wu, Min; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Artificial intelligence (AI) is being used across various practical tasks in recent years, facilitating many aspects of our daily life. With AI-based sensing and learning systems, we can enjoy the services of automated decision making, computer-assisted medical diagnosis, and health monitoring. Since these algorithms have entered human society and are influencing our daily life, such important issues as intellectual property protection, access control, privacy protection, and fairness/equity, should be considered when we are developing the algorithms, in addition to their successful performance. In this dissertation, we improve the design of emerging AI-based sensing and learning systems from security and fairness perspectives. The first part is the security protection of deep neural networks (DNN). DNNs are becoming an emerging form of intellectual property for model owners and should be protected from unauthorized access and piracy to encourage healthy business investment and competition. Taking advantage of DNN's intrinsic mechanism, we propose a novel framework to provide access control to the trained DNNs so that only authorized users can utilize them properly to prevent piracy and illicit usage. The second part is privacy protection in facial videos. Remote Photoplethysmography (rPPG) can be used to collect a person's physiological signal when his/her face is captured by a video camera, which may raise privacy issues from two aspects. First, individual health conditions may be revealed from a facial recording unintentionally by a person without his/her explicit consent from a facial recording. To avoid the physiological privacy issue, we develop \textit{PulseEdit}, a novel and efficient algorithm that can edit the physiological signals in facial videos without affecting visual appearance to protect the person's physiological signal from disclosure.On the other hand, R\&D of rPPG technology also has a potential leakage of identity privacy. We usually require public benchmark facial datasets to develop rPPG algorithms, but facial videos are often very sensitive and have a high leakage risk in identity privacy. We develop an anonymization transform that removes sensitive visual information identifying an individual, but in the meantime, preserves the physiological information for rPPG analysis. In the last part, we investigate fairness in machine learning inference. Various fairness definitions in prior art were proposed to ensure that decisions guided by the machine learning models are equitable. Unfortunately, the ``fair'' model trained with these fairness definitions is sensitive to threshold, i.e., the condition of fairness will no longer hold when tuning the decision threshold. To this end, we introduce the notion of threshold-invariant fairness, which enforces equitable performances across different groups independent of the decision threshold.Item On agent-based modeling: Multidimensional travel behavioral theory, procedural models and simulation-based applications(2015) Xiong, Chenfeng; Zhang, Lei; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation proposes a theoretical framework to modeling multidimensional travel behavior based on artificially intelligent agents, search theory, procedural (dynamic) models, and bounded rationality. For decades, despite the number of heuristic explanations for different results, the fact that "almost no mathematical theory exists which explains the results of the simulations" remains as one of the large drawbacks of agent-based computational process approach. This is partly the side effect of its special feature that "no analytical functions are required". Among the rapidly growing literature devoted to the departure from rational behavior assumptions, this dissertation makes effort to embed a sound theoretical foundation for computational process approach and agent-based microsimulations for transportation system modeling and analyses. The theoretical contribution is three-fold: (1) It theorizes multidimensional knowledge updating, search start/stopping criteria, and search/decision heuristics. These components are formulated or empirically modeled and integrated in a unified and coherent approach. (2) Procedural and dynamic agent-based decision-making is modeled. Within the model, agents make decisions. They also make decisions on how and when to make those decisions. (3) Replace conventional user equilibrium with a dynamic behavioral user equilibrium (BUE). Search start/stop criteria is defined in the way that the modeling process should eventually lead to a steady state that is structurally different to user equilibrium (UE) or dynamic user equilibrium (DUE). The theory is supported by empirical observations and the derived quantitative models are tested by agent-based simulation on a demonstration network. The model in its current form incorporates short-term behavioral dimensions: travel mode, departure time, pre-trip routing, and en-route diversion. Based on research needs and data availability, other dimensions can be added to the framework. The proposed model is successfully integrated with a dynamic traffic simulator (i.e. DTALite, a light-weight dynamic traffic assignment and simulation engine) and then applied to a mid-size study area in White Flint, Maryland. Results obtained from the integration corroborate the behavioral richness, computational efficiency, and convergence property of the proposed theoretical framework. The model is then applied to a number of applications in transportation planning, operations, and optimization, which highlights the capabilities of the proposed theory in estimating rich behavioral dynamics and the potential of large-scale implementation. Future research should experiment the integration with activity-based models, land-use development, energy consumption estimators, etc. to fully develop the potential of the agent-based model.