2020 Theses (MU)https://hdl.handle.net/10355/779452024-03-29T10:46:34Z2024-03-29T10:46:34ZAdversarial robustness of deep learning enabled industry 4.0 prognosticsMode, Gautam Rajhttps://hdl.handle.net/10355/889522023-12-14T18:54:14Z2020-01-01T00:00:00ZAdversarial robustness of deep learning enabled industry 4.0 prognostics
Mode, Gautam Raj
The advent of Industry 4.0 in automation and data exchange leads us toward a constant evolution in smart manufacturing environments, including extensive utilization of Internet-of-Things (IoT) and Deep Learning (DL). Specifically, the state-of-the-art Prognostics and Health Management (PHM) has shown great success in achieving a competitive edge in Industry 4.0 by reducing maintenance cost, downtime, and increasing productivity by making data-driven informed decisions. These state-of-the-art PHM systems employ IoT device data and DL algorithms to make informed decisions/predictions of Remaining Useful Life (RUL). Unfortunately, IoT sensors and DL algorithms, both are prone to cyber-attacks. For instance, deep learning algorithms are known for their susceptibility to adversarial examples. Such adversarial attacks have been extensively studied in the computer vision domain. However, it is surprising that their impact on the PHM domain is yet not explored. Thus, modern data-driven intelligent PHM systems pose a significant threat to safety- and cost-critical applications. Towards this, in this thesis, we propose a methodology to design adversarially robust PHM systems by analyzing the effect of different types of adversarial attacks on several DL enabled PHM models. More specifically, we craft adversarial attacks using Fast Gradient Sign Method (FGSM) and Basic Iterative Method (BIM) and evaluate their impact on Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN), Bi-directional LSTM, and Multi-layer perceptron (MLP) based PHM models using the proposed methodology. The obtained results using NASA's turbofan engine, and a well-known battery PHM dataset show that these systems are vulnerable to adversarial attacks and can cause a serious defect in the RUL prediction. We also analyze the impact of adversarial training using the proposed methodology to enhance the adversarial robustness of the PHM systems. The obtained results show that adversarial training is successful in significantly improvising the robustness of these PHM models.
2020-01-01T00:00:00ZAnchor team gender make-up and audience evaluationNinke, Kathleen Wittehttps://hdl.handle.net/10355/839312023-12-14T18:53:54Z2020-01-01T00:00:00ZAnchor team gender make-up and audience evaluation
Ninke, Kathleen Witte
For decades, the man-woman anchor team has been the industry standard in both national and local news. Recently, high-profile national newscasts have used two anchors of the same gender to headline the broadcast. In local news, while scientific data is not available, the standard seems to remain: newsroom managers generally hire for a man-woman newscast. However, per this experimental study, audiences do not prefer the man-woman team over the man-man or woman-woman team with any statistical significance. Furthermore, audiences rated a two-woman team more "credible" than its counterparts. This research explores the heuristics and socialization that may inform an audience member's evaluation of a newscast and its anchor team while questioning the mixed-gender standard.
2020-01-01T00:00:00ZAnisotropy and mantle flow of the Indo-Burma subduction zone from shear wave splitting and shear wave splitting tomographyIslam, Md Mohimanulhttps://hdl.handle.net/10355/889452023-12-14T18:54:13Z2020-01-01T00:00:00ZAnisotropy and mantle flow of the Indo-Burma subduction zone from shear wave splitting and shear wave splitting tomography
Islam, Md Mohimanul
The Indo-Burma subduction zone lies on the eastern boundary of the 1500 km long Alpine-Himalayan orogenic belt. Despite highly oblique plate motion with limited arc development, convergence is still ongoing with the potential to generate megathrust earthquakes that could affect more than 140 million people. This sub-aerial subduction system provides an excellent opportunity to study the nature of mantle deformational fabric resulting from the active convergence and collision with possible clockwise rotation of the Indian plate. Characterizing the mantle flow field helps us to better understanding the geodynamics and regional tectonics of this part of the plate boundary. Shear wave splitting is a simple yet powerful technique to investigate anisotropy and mantle strain fabric. I have used both teleseismic and local shear waves to create a detailed map of the upper mantle anisotropic fabrics using a recently deployed temporary and permanent seismic stations, across the northern and central parts of Myanmar. Along with the detailed analysis of splitting parameters, I have generated a tomographic model inverting local and teleseismic shear waves splitting results. The resulting 3D anisotropic model helps to image the prevailing mantle flow below the slab in the sub-slab region and above the slab in the mantle wedge. I found a trench parallel fast direction with high lag time (>2.0s) on the fore-arc that primarily accumulated in the sub-slab region. Such observed fabric results from the toroidal flow around the slab in response to the slab roll-back. The mantle wedge shows a complex deformational fabric with an average lag time of over 0.5s. Dextral motion along the Sagaing fault induces a dominant trench parallel fabric on the mantle wedge. Null results in the southern part of the Shan plateau suggest that back-arc upwelling may exist between the transition of the east-west fast direction on the southern Yunnan and the north-south fast direction of the western subduction margin. The nature of the transition of the mantle fabric and back-arc upwelling is not well understood due to the lack of seismic station coverage on the Shan plateau.
2020-01-01T00:00:00ZAnnual evidence of moisture limitations at treeline in the Sangre de Cristo MountainsBailey, Sydney N.https://hdl.handle.net/10355/785812023-12-14T18:53:54Z2020-01-01T00:00:00ZAnnual evidence of moisture limitations at treeline in the Sangre de Cristo Mountains
Bailey, Sydney N.
This study is focused on capturing a recent regeneration patterns on an annual level over the landscape of the Sangre de Cristo (SDC) mountain range. Modern temperature trends (post-1945) of this area are dominated by sharp rises in minimum temperature during the warm and cool seasons. Coincidentally, the onset of heat-induced drought stress is impacting trees throughout the mountain forest belt, though research is lacking across broad spatial scales at treeline. I used dendroecological techniques to destructively sample seedlings on contrasting north and south facing slopes at upper treeline in the SDC to investigate moisture interactions at treeline. I hypothesized that if patterns of tree regeneration are primarily driven by temperature, then I would expect seedling establishment to be more abundant on south-facing slopes. Alternatively, if heat induced drought stress is an important driver, I would expect seedling establishment to be more confined to north-facing slopes. Results show that seedling establishment is significantly (p < 0.01) favored on north-facing slopes (n=169) verses south-facing slopes (n = 66). These seedlings are also significantly younger (p < 0.05) and smaller (p < 0.01) than their counterparts on south-facing slopes. The relationship between climate and annual establishment patters were investigated to identify any important drivers. Results show that there is a negative relationship between drought conditions and establishment events. Across all sites (n=6) and slope aspects, there was no establishment found after 2009, possibly indicating a minimum threshold has been surpassed and conditions are no longer suitable to regeneration above treeline. This indicates the possibility of moisture limitations at treeline and brings into question the future structure and extent of the upper forest border under a warmer and drier climate.
2020-01-01T00:00:00Z