教学课程资源库 更多>>

关键词： Mechanical engineering   Computational physics   Computer science   Artificial intelligence   Mechanics   Industrial engineering  

摘要： Manufacturing processes are known for their intricacies in changing material shapes and properties. New generations of manufacturing technologies, known as flexible manufacturing, are moving toward design freedom, which allows producing parts with optimized geometries and high customizations at an affordable cost even for low-volume productions. Two prominent flexible manufacturing processes that are of interest in this dissertation are additive manufacturing and incremental sheet forming. An important limiting factor in advancing current capabilities in such processes is the difficulty to reliably understand and control them due to the complex multi-physics and multi-scale nature of the processes. As the result, current practices are overly conservative, significantly limiting the vast potential of producing parts with customized material properties. At the same time, we observe a surge in the digitalization of manufacturing processes. Today, manufacturing facilities are more connected to data centers than ever, and various measurement methods are becoming standard components of modern manufacturing pipelines from controlling and monitoring the progress while manufacturing to testing and analysis after the part is built. Therefore, this dissertation is dedicated to developing computational methods to advance modeling and design capabilities with a focus on approaches to optimally use manufacturing data—an underutilized asset of manufacturing systems. My contributions in process characterization and design are organized into five research tasks and briefly discussed below. Predicting the spatiotemporal behavior of manufacturing processes is challenging due to the long history-dependent correlations and complex unstructured geometric features common in manufacturing. Motivated by this challenge, my first research contribution introduces a data-driven methodology to learn material behaviors on unseen geometries over long simulation periods. My method efficiently combi

关键词： Immunology   Biology   Morphology   Cellular biology   Computer science   Medicine   Physiology  

摘要： In addition to providing a physical barrier, the epidermis provides a niche for the long-term residency of Langerhans cells (LC). LC provide protection by priming elements of the adaptive immune system, but they can also participate in the maintenance and development of autoimmune and inflammatory diseases. Within the epidermis, LC possess the ability to migrate from the skin epidermis to regional lymph nodes (LN). At steady-state, LC migrate to LN transporting self and potentially commensal-derived antigen to promote the development of peripheral tolerance. During infection, LC transport potential pathogen-derived antigens and prime naïve adaptive immune responses. The mechanisms that allow for the long-term retention of these leukocytes has not been extensively studied. Hence, elucidating the mechanisms by which LC maintain their epidermal residency and mediate migration is of great therapeutic interest and will extend our knowledge in LC biology. Epidermal integrins ανβ6 and ανβ8 on keratinocytes (KCs) cleave LAP-TGFβ1 to its active form TGFβ1which acts on the TGFβ receptor of LCs to maintain their long-term residency within the skin epidermis. We have also demonstrated that selective expression of a ligand independent constitutively active form of TGFβRI, inhibits LC migration during homeostasis and in response to UVB exposure. Interestingly, UVB treatment resulted in the loss of epidermal integrins ανβ6 and ανβ8 on KCs. However, whether inflammation induced LC migration acts through a similar mechanism remains unclear. Thus, we propose to test the hypothesis that inflammatory stimuli mediate LC migration from the epidermis through the inhibition of KC-integrin mediated TGF-β activation. Here we demonstrate that LC migration in response to inflammatory stimuli was also inhibited by ligand independent TGFβRI signaling. Unlike UVB stimulation which reduced KCs expression of avβ6, in vitro and in vivo exposure to TNF-α or IL-1β increased avβ6 mRNA and protein expre

关键词： Computer science   Aerospace engineering   Robotics  

摘要： Real-time object classification, localization, and detection with region-based convolution neural network (R-CNN) require high computational power, or it consumes a tremendous amount of time with the use of the available onboard computer system. Therefore, either of those ways is not practical in real-time object detection. In contrast, the YOLO-v3 network, which stands for you only look once, that uses the YOLO algorithm seems to be practical in live object classification, localization, and detection, since the YOLO algorithm seems to work faster than the sliding window algorithm used by R-CNN. Sensor fusion in this context requires estimation and association of information coming from more than one optical sensor that improves the accuracy in locating the object in the real world. The goal of this research is to implement a real-time object detection and tracking algorithm for a collision avoidance application in an autonomous ground vehicle. Transfer learning was performed to train the pre-trained artificial neural network (YOLO-v3) that uses a supervised learning algorithm, and a labeled dataset was created to train the deep neural network. Hyper-parameters such as mini-batch size, L2 regularization parameter, number of epochs, and learning rate were tuned to study the trade-off between the bias and variance of a classifier to improve the performance of the neural network. Moreover, the intrinsic calibration of the stereo camera and LiDAR was performed to find standard measurement error by re-projecting the measurement data on ground truth data. Similarly, the extrinsic calibration was done to derive standard transformation error between these two optical sensors. Finally, the object's states were fed to the centralized sensor fusion framework, which uses a linear Kalman filter, to estimate the states of the objects. The complete framework was developed in MATLAB 2020b by use of deep learning, computer vision and image processing, LiDAR, and sensor fusion and tr

关键词： Computer science   Computer engineering   Engineering  

摘要： Knowing the expert developers of a software development project has great significance in large-scale and geographically distributed projects. However, finding these expert developers can be challenging, which becomes more complicated over time as the development team gets bigger and more distributed. This thesis presents an expert developer recommender system for methods, based on the usage expertise, implementation expertise, and the combination of both, for the developers of a software project. A developer acquires usage expertise on a method by calling or using it and implementation expertise by creating or modifying it. To determine the method expertise of the developers, we mine both the source code repository and the version histories of a software development project. We determine the accuracy of our system by calculating the percentage of successful recommendations within the Top-N results. Through several experiments, we found that our recommender system produces around 90% average accuracy for Top-10 recommendations.

关键词： Computer Engineering   Information Technology   Computer science   Artificial intelligence  

摘要： Deep learning or deep neural networks (DNNs) have achieved extraordinary performance in many application domains such as image classification, object detection and recognition, natural language processing and medical image analysis. It has been well accepted that DNNs are vulnerable to adversarial attacks, which raises concerns of DNNs in security-critical applications and may result in disastrous consequences. Adversarial attacks are usually implemented by generating adversarial examples, i.e., adding sophisticated perturbations onto benign examples, such that adversarial examples are classified by the DNN as target (wrong) labels instead of the correct labels of the benign examples. The adversarial machine learning aims to study this phenomenon and leverage it to build robust machine learning systems and explain DNNs. In this dissertation, we present the mechanism of adversarial machine learning in both empirical and theoretical ways. Specifically, we first introduce a uniform adversarial attack generation framework, structured attack (StrAttack), which explores group sparsity in adversarial perturbations by sliding a mask through images aiming for extracting key spatial structures. Second, we discuss the feasibility of adversarial attack in the physical world and introduce a powerful framework, Expectation over Transformation (EoT). Utilize EoT with Thin Plate Spline (TPS) transformation, we can generate Adversarial T-shirts, a robust physical adversarial example for evading person detectors even if it could undergo non-rigid deformation due to a moving person’s pose changes. Third, we stand on the defense side and propose the first adversarial training method based on Graph Neural Network. Fourth, we introduce Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes the provable linear bounds of output neurons given a certain amount of input perturbation. LiRPA studies the adversarial example in a theoretical way and can guarante

关键词： Computer science   Artificial intelligence  

摘要： State-of-the-art conversational agents have advanced significantly in conjunction with the use of large transformer-based language models. However, even with these rapid advancements, the current generation of conversational agents suffers from three major problems: (i) long-term context modeling; (ii) producing informative and factually accurate responses; (iii) robust evaluation of the NLG systems. Our work tackles these three gaps: (i) To address the issue of long-term context modeling, we present a novel end-to-end approach inspired by neurocognitive memory processes. We also implement a novel action selection mechanism that helps identify the relevant utterances containing salient information from long-term memory to working memory to better incorporate the context of the conversation during the generation process than state-of-the-art systems. (ii) To integrate knowledge into conversational agents, we also propose a dialog framework that incorporates both local knowledge as well as users' past dialogues to generate high-quality personalized conversations. Using our framework, we demonstrate that incorporating local knowledge can largely improve informativeness, coherency and realisticness measures using human evaluations. However, even with these advancements, we find that knowledge grounded conversation models are prone to hallucinations. We address this issue by proposing a new dataset called "CONV-FEVER'' to build a fact consistency detector. We show that our detector outperforms the current SOTA and can be integrated with existing models to increase the factual consistency of the knowledge grounded models. (iii) In the last part of this thesis, we focus on the aspect of the impact of experiment design in conversational AI systems by conducting two large-scale studies. In the first study, we compare four different experimental designs and study how each experiment design affects the quality of outputs obtained from the human evaluation. In the second study,

关键词： Computer science   Artificial intelligence  

关键词： Software   Zoos   Marine mammals   Births   Mitochondrial DNA   Wildlife conservation   Animal sciences   Computer science   Wildlife management  

关键词： Artificial intelligence   Computer science   Electrical engineering  

摘要： Machine learning is entering every aspect of our lives, including high-stakes applications that directly affect people's lives, such as, hiring, education, lending, or healthcare. While these machine learning models are undoubtedly great at learning all patterns present in the data, blindly learning all patterns can often have unintended consequences, such as propagating biases with respect to gender, race, etc. These biases can adversely affect people’s lives, and also violate anti-discrimination laws, e.g., Title VII of the US Civil Rights Act. When it comes to resolving legal disputes, or even informing policies and interventions, only identifying bias and disparity in a model’s decision is not always sufficient. We really need to dig deeper and be able to identify/explain the sources of disparity. For example, disparities in hiring that can be explained by an occupational necessity may be exempt by law, e.g., code-writing skills for hiring a software engineer for a safety-critical application. However, the disparity arising due to an aptitude test may not be exempt, as in the landmark court case of Griggs v. Duke Power ‘71. This leads us to a question that bridges the fields of fairness, explainability, and law: How can we identify and explain the sources of disparity in machine learning models, e.g., did the disparity arise solely due to the critical occupational necessities? In this dissertation, I propose a systematic measure of "non-exempt disparity,'' i.e., the disparity which cannot be accounted for by the occupational necessities. To arrive at a measure for the non-exempt disparity, I adopt a rigorous axiomatic approach that brings together concepts in information theory, in particular, an emerging body of work called Partial Information Decomposition, with Pearl's causality. This dissertation also examines an extension of this technique to quantifying contribution of each individual feature to the observed disparity, a novel form of explainability. Lastl

关键词： Computer science   Artificial intelligence  

摘要： Massive Online Open Courses (MOOCs) have become increasingly popular among students, partly because they make learning resources much easier to spread and share. However, MOOC's unfinished rate or failure rate can be extremely high. To improve the current state of affairs, were sort to learning and discovering students' online study behavior. By using the learning behavior features that we collect, we first build a statistical model that predicts a student's examination scores. Furthermore, at a deeper level, we extract language features from students' reading notes. Through analyzing the topics in the reading notes, we obtain latent but robust and reliable features that indicate a student's performance in the course. Specifically, we use a Topic-based Latent Variable Model to predict the midterm and final scores with students' textbook reading notes, and compare with both the Two-step LDA model and the knowledge graph embedding. Our novel study brings insights into understanding the the connection between online course study behaviors and end performance, which will not only benefit the conscientious students who want to improve their study, but also shed light on educators for identifying the under-performing students earlier for intervention, and for improving the design of online courses. Last but not least, we study the theoretical problem and prove the convergence of prediction accuracy for a generic discriminative model.