教学课程资源库 更多>>

关键词： Computer engineering   Artificial intelligence   Computer science  

摘要： Due to inherent complex behaviors and stringent requirements in analog and mixed-signal (AMS) systems, verification and testing become key bottlenecks in the product development cycle. Rare failure detection in a high-dimensional parameter space using minimal expensive simulation/measurement data is a major challenge. For rare failure detection in the verification flow, this dissertation proposes to put machine learning models, that mimic the circuit behavior, under verification, which greatly relaxes the simulation/measurement requirements and improves the verification efficiency. We first present a hybrid formal/machine-learning verification technique (HFMV) to combine the best of the two worlds. HFMV adds formalism on the top of a probabilistic learning model while providing a sense of coverage for extremely rare failure detection. On the other hand, we also study Bayesian optimization (BO) based approaches to the challenging problem of verifying AMS circuits with stringent low failure requirements. We simultaneously leverage multiple optimized acquisition functions to explore varying degrees of balancing between exploitation and exploration. Furthermore, this dissertation proposes a BO framework under high dimensional space to further improve the verification efficiency. Two techniques are explored here: 1) random embedding to linearly embed input into a low dimensional space and 2) sensor fusion networks to identify important nonlinear features transformed by reversible neural networks. The proposed approaches are very effective in finding very rare failures in high dimensional space which existing statistical techniques can miss. On the subject of AMS testing, this dissertation proposes to utilize self-supervised learning methods to detect extremely rare customer failure. First, we study a transformation-based self-labeling technique to reliably screen out rare customer return defects. The normality score to an unseen input data is the goodness of the multi-cl

关键词： Applied mathematics   Mathematics   Computer science  

摘要： We propose model-based derivative-free trust-region algorithms for constrained optimization problems involving black-box functions. Our algorithms are specifically focused on problems for which infeasible points do not provide function values. The algorithms work by approximating the objective function and each constraint function by polynomial model functions constructed by interpolating over a feasible set of evaluated sample points. The fact that function values cannot be obtained outside of the feasible region mandates that only feasible sample points can be used to construct the model functions. This restriction poses interesting challenges for ensuring that the sample set is well-poised, which means that relative positions of the sample points ensure that the interpolated model functions will be sufficiently accurate approximations of the true functions. To address these challenges, we propose two strategies for constructing sample sets that ensure good geometry for the sample set: an ellipsoidal trust region strategy and a polyhedral trust region strategy. Chapter 3 of the thesis is focused on the special case of linear constraints. For this case, we present several methods for constructing ellipsoidal sample regions within the intersection of the feasible region and an outer trust region. By adapting a well-known model-improvement algorithm for selecting sample points, we show how to construct a well-poised sample set over this ellipsoidal region. This result is used to show that under reasonable assumptions, the iterates produced by our trust region algorithm have criticality measures that converge to zero. This guarantees, in particular, that any accumulation point of the iterates satisfies the first order optimality conditions. In addition to the ellipsoidal trust region strategy, we also develop a polyhedral trust region strategy. For this strategy, we propose a modification to the model improvement algorithm aimed at constructing a well-poised sample se

关键词： Computer science   Operations research   Web studies  

关键词： Computer science   Ophthalmology  

摘要： Colour Vision Deficiency (CVD) is an important issue for a significant population across the globe. There are several types of CVD's, such as monochromacy, dichromacy, trichromacy, and anomalous trichromacy. Each of these categories contain specific other subtypes. The aim of this research is to device a scheme to address CVD by using variations in pixel plotting of colours to capture colour disparities and perform colour compensation. The proposed scheme recolours the video and images by colour contrast variation of each colour for CVD patients, and depending on the type of deficiency, it is able to provide live results. Different types of CVD’s can be identified and cured by changing the particular colour related to it and based upon the type of diseases, it performs RGB (Red, Green, and Blue) to LMS (Long, Medium, and Short) transformation. This helps in colour identification and also adjustments of colour contrasts. The processing and rendering of recoloured video and images, allows the affected patients with CVD to see perfect shades in the recoloured frames of video or images and other modes of files. In this thesis, we propose an efficient recolouring algorithm with a strong focus on real-time applications that is capable of providing different recoloured outputs based on specific types of CVD.

关键词： Computer science   Public health   Web studies   Pathology  

摘要： SARS CoV-2 (COVID-19) was identified as the cause of severe respiratory disease in China in 2019. It is a virus that will be transferred person-to-person by sneezing, coughing, or talking. This phenomenon not only affects public health and economics but also mental health as well. SARS-CoV-2 vaccines and wearing masks plays significant roles in preventing the spread of the COVID-19 virus, but vaccine hesitancy and anti-mask beliefs threaten the efficacy of the government orders in prevention and immunization against Coronavirus. The impact of the COVID-19 pandemic has been investigated from different aspects, but few large-scale studies focus on the opinion of people toward government orders to wear face mask and get vaccination. The abundant data on online social media however enables researchers to analyze people's attitudes toward vaccination and the use of face mask. In this study, we use Twitter API and scrape 340 million COVID-19 tweets posted in the timeline of December 2020 to March 2021. Our goal is to investigate how people respond to tweets about masking and vaccines as a means of understanding sentiments towards both practices. Specifically, we focus on which tweets about the topics tend to become viral relative to those that are neither retweeted nor receive any replies. Toward this end, we split the dataset into three categories: 1) replied tweets 2) retweeted tweets, and 3) no-engagement tweets which are tweets that receive no response. We then deploy topic modeling to identify the most popular tweet topics in each category. Furthermore, we filter tweets for vaccine and mask related hashtags and use the algorithm, VADER to find the sentiment of these tweets. By applying topic modeling and Vader, we assess the vaccine and mask-related sentiment scores and visualize their progression during four months. Our analysis indicates a slight difference in the distribution of tweets with positive and negative sentiments with vaccination or mask hashtags, with th

关键词： Computer science   Artificial intelligence  

关键词： Operations research   Computer science   Information technology  

摘要： The purpose of this research project was to identify cybersecurity vulnerabilities in automating industrial control systems (ICS) or supervisory control and data acquisition (SCADA) systems. Generally, there is a lack of cybersecurity plans in automating ICS or SCADA systems. This project intended to answer three questions. 1. How have supply chains been compromised, affecting international automated ICS or SCADA systems? 2. Threats posed by automated ICS or SCADA systems to the confidentiality, integrity, and availability (CIA) triad are significant, and how are they enumerated? 3. How can policies be improved within the automated ICS or SCADA systems community to prevent further supply chain cyberattacks? This research project is for cybersecurity professionals, landlords and operators of ICS or SCADA systems, and owners and government leaders of critical infrastructures. The research utilized for this project includes peer-reviewed journals, research projects, and published information from the International Electrotechnical Commission (IEC), an international standards organization for all electrical, electronic, and related technologies. Keywords: Cybersecurity, Professor Carmen Mercado, intrusion detection system (IDS), Internet of Things (IoT), programmable logic controller (PLC), smart grid (SG), cyberattacks.

关键词： Logic gates   Object detection   Contamination   Task analysis   Saliency detection   Computer science   Image color analysis   Salient object detection   RGB-D images   depth potentiality perception   gated multi-modality attention  

摘要： There are two main issues in RGB-D salient object detection: (1) how to effectively integrate the complementarity from the cross-modal RGB-D data;(2) how to prevent the contamination effect from the unreliable depth map. In fact, these two problems are linked and intertwined, but the previous methods tend to focus only on the first problem and ignore the consideration of depth map quality, which may yield the model fall into the sub-optimal state. In this paper, we address these two issues in a holistic model synergistically, and propose a novel network named DPANet to explicitly model the potentiality of the depth map and effectively integrate the cross-modal complementarity. By introducing the depth potentiality perception, the network can perceive the potentiality of depth information in a learning-based manner, and guide the fusion process of two modal data to prevent the contamination occurred. The gated multi-modality attention module in the fusion process exploits the attention mechanism with a gate controller to capture long-range dependencies from a cross-modal perspective. Experimental results compared with 16 state-of-the-art methods on 8 datasets demonstrate the validity of the proposed approach both quantitatively and qualitatively. https://***/JosephChenHub/DPANet

关键词： Computer engineering   Computer science   Artificial intelligence  

摘要： Detecting whether or not a software is malicious is a difficult task. The context under which software is executing plays an important role in understanding malicious behavior. Current malware detection strategies have shown high classification accuracy, but they lack contextual considerations. The objective of this thesis is to address the development of a context-aware malware detection system. A definition of context and how it pertains to malware detection is discussed. Based on this definition, two proof-of-concept context-aware models utilizing Latent Dirichlet Allocation are developed to address different aspects of context. These models provide insight into the challenges of including context in malware detection models, and future work to improve the contextual aspects of the models is discussed.

关键词： Mechanical engineering   Computer science   Thermodynamics   Automotive engineering