教学课程资源库 更多>>

关键词： Computer science   Electrical engineering   Electromagnetics  

摘要： Brushless DC (BLDC) motors are becoming increasingly common in various industrial and commercial applications such as micromobility and robotics due to their high torque density and efficiency. A BLDC Motor is a three-phase synchronous motor that is very similar to a non-salient Permanent Magnet Synchronous Motor (PMSM) with key differences lying in the non-ideal characteristics of the motor; the most prominent of these is BLDC motors have trapezoidal-shaped Back-Electromotive Force (BEMF). Despite their advantages, a present weakness of BLDC motors is the difficulty controlling these motors at standstill and low-speed conditions that require high torque. These operating conditions are common in the target applications and almost always necessitate the use of external sensors which introduce additional costs and points of failure. As such, sensorless based methods of position estimation would serve to improve system reliability, cost, and efficiency. High Frequency (HF) pulsating voltage injection in the direct axis is a popular method of sensorless control of salient-pole Interior-mount Permanent Magnet Synchronous Motors (IPMSM); however, existing methods are not sufficiently robust for use in BLDC and small Surface-mount Permanent Magnet Synchronous Motors (SPMSM) and are accompanied by other issues, such as acoustic noise. This thesis proposes novel improvements to the method of High Frequency Voltage Injection to allow for practical use in BLDC Motors and small SPMSM. Proposed improvements include 1) a hybrid frequency generator which allows for dynamic frequency scaling to improve tracking and eliminate acoustic noise, 2) robust error calculation that is stable despite the non-ideal characteristics of BLDC Motors, 3) gain engineering of Proportional-Integral (PI) type Phase-Locked-Loop (PLL) trackers that further lend stability, 4) observer decoupling mechanism to allow for seamless transition into state-of-the-art BEMF sensing methods at high speed, and 5) sa

关键词： Computer science   Theoretical mathematics   Mathematics  

关键词： Artificial intelligence   Computer science   Software   Datasets   Coronaviruses   Social network analysis  

摘要： Social Network Analysis (SNA) is an appealing research topic, within the domain of Artificial Intelligence (AI), owing to its widespread application in the real world. In this dissertation, we have proposed effective Machine Learning (ML) and Deep Learning (DL) approaches toward resolving these open problems with regard to SNA, viz: Breakup Prediction, Link Prediction, Node Classification, Event-based Analysis, and Trend/Pattern Analysis. SNA can be employed toward resolving several real-world problems; and ML as well as DL have proven to be very effective methodologies for accomplishing Artificial Intelligence (AI)- related goals. Existing literature have focused on studying the apparent and latent interactions within social graphs as an n-ary operation, which yields binary outputs comprising positives (friends, likes, etc.) and negatives (foes, dislikes, etc.). Inasmuch as interactions constitute the bedrock of any given Social Network (SN) structure; there exist scenarios where an interaction, which was once considered a positive, transmutes into a negative as a result of one or more indicators which have affected the interaction quality. At present, this transmutation has to be manually executed by the affected actors in the SN. These manual transmutations can be quite inefficient, ineffective, and a mishap might have been incurred by the constituent actors and the SN structure prior to a resolution. Thus, as part of the research contributions of this dissertation, we have proposed an automatic technique toward flagging positive ties that should be considered for breakups or rifts (negative-tie state), as they tend to pose potential threats to actors and the SN. Furthermore, in this dissertation, we have proposed DL-based approaches based on edge sampling strategy for resolving the problems of Breakup Prediction, Link Prediction, and Node Classification. Also, we have proposed ML-based approaches for resolving the problems of Event-based Analysis and Trend/Patte

关键词： Engineering   Computer science   Automotive engineering   Transportation   Robotics  

摘要： Growing Electric Vehicle penetration presents unwanted problems to grid reliability. For nearly every EV there is corresponding household charging. High penetration of Electric Vehicle Service Equipment can lead to over loading of assets and under voltage conditions. In order to understand the effects EV have on a distribution system, studies have to be done for EVSE to understand how the distribution system is affected. Using advanced Power Engineering Simulation Software is often the best way to model systems due to the credibility of their software modules. For this thesis, I developed a suite of distribution system analysis tools using CYME 7.1and Python 2.7 for evaluating the impacts from EV penetration, particularly overloading and under-voltage events. EV penetration is the percentage of electric vehicles among total vehicles. Two of these tools apply new loads to and create intentional spot loads on the provided system. Another two tools incorporate time series demands for EV loads and provide load growth on the system. The final tool covers data collection for over loading and under voltage events. Through use of this work’s EV Evaluation Tools, users can study how a distribution system may be impacted due to EV load growth and stochastic EV placement. These tools allow for a representation of how the system changes with increased EV penetration, at the years the penetration are projected to increase.

关键词： Artificial intelligence   Computer science  

摘要： Mental or emotional states are an important part of the information that can be observed from the speakers or writers of many discoures. On social media, people often share their emotional responses to events, stories, news, etc. The fundamental goal of this dissertation is to identify and analyze various types of mental or emotional states. The freedom of accessibility and speech that social media provides, allows people to express their feelings and beliefs, making it possible to judge users' personality and mental state. Traditional methods to identify mental health issues rely on self-reported methods in well-controlled, decontextualized environments. However, social media can be used as a new lens to explore and understand peoples' behavior more accurately in daily basis routine, where they share and exchange their feeling, opinions, plans, and even some private information such as medical diagnosis. Different transformer-based models are utilized to predict these types of mental states on social media content. Accurately modeling mental health issues in social media has many important applications to personal health individually and society wellness collectively. In this dissertation, I propose different transformer-based models to predict these types of mental states on social media content. Building models that can capture left and right contexts jointly in sentences are necessary to enrich word representation and this is why transformer-based models like BERT and XLNet are used in this study. Utilizing multiple-head self attentions via multi-layer transformers, these models are able to model negations and other semantic relationships by paying attention to crucial words, leading to more accurate predictive models for optimism and pessimism, and suicide. It is wildly recognized and observed that deep models suffer from issues like overfitting phenomena which preventing the model from generalizing to new data. We propose a novel method called Soft Label Assig

关键词： Civil engineering   Design   Architecture   Computer science   Engineering   Mathematics   Mechanics   Physics  

关键词： Computer science   Artificial intelligence  

摘要： Face recognition is a widely adopted technology with numerous applications, such as mobile phone unlock, mobile payment, surveillance, social media and law enforcement. There has been tremendous progress in enhancing the accuracy of face recognition systems over the past few decades, much of which can be attributed to deep learning. Despite this progress, several fundamental problems in face recognition still remain unsolved. These problems include finding a salient representation, estimating intrinsic dimensionality, representation capacity, and demographic bias. With growing applications of face recognition, the need for an accurate, robust, compact and fair representation is evident. In this thesis, we first develop algorithms to obtain practical estimates of intrinsic dimensionality of face representations, and propose a new dimensionality reduction method to project feature vectors from ambient space to intrinsic space. Based on the study in intrinsic dimensionality, we then estimate capacity of face representation, casting the face capacity estimation problem under the information theoretic framework of capacity of a Gaussian noise channel. Numerical experiments on unconstrained faces (IJB-C) provide a capacity upper bound of 27,000 for FaceNet and 84,000 for SphereFace representation at 1% FAR. In the second part of the thesis, we address the demographic bias problem in face recognition systems where errors are lower on certain cohorts belonging to specific demographic groups. We propose two de-biasing frameworks that extract feature representations to improve fairness in face recognition. Experiments on benchmark face datasets (RFW, LFW, IJB-A, and IJB-C) show that our approaches are able to mitigate face recognition bias on various demographic groups (biasness drops from 6.83 to 5.07) as well as maintain the competitive performance (i.e., 99.75% on LFW, and 93.70% TAR @ 0.1% FAR on IJB-C). Lastly, we explore the global distribution of deep face representati

关键词： Game theory   Algorithms   Decision trees   Success   Cooperation   Attitudes   Decision making   Neural networks   Artificial intelligence   Computer science  

摘要： As Artificial Intelligence systems are used by human users at an increasing frequency, the need for such systems to understand and predict human behavior likewise increases. In my work, I have considered how to predict human behavior in repeated games. These repeated games can be applied as a foundation to many situations where a person may interact with an AI, In an attempt to create such a foundation, I have built a system using Attitude Vectors used in automata to predict actions based on prior actions and communications. These Attitude Vector Automata (AVA) can transform information from actions in one game with a given payoff matrix into actions in another game. Results show that prediction accuracy was ultimately below other, similar work, in general in several repeated games. There are however some aspects, such as scenarios involving lying, in which my predictor showed potential to outperform these other systems. Ultimately, there is potential in using ideas presented as AVA to build a potentially more robust system for future efforts in human behavior prediction.

关键词： Computer science   Artificial intelligence   Bioinformatics  

关键词： Electrical engineering   Computer science   Artificial intelligence   Applied mathematics  

摘要： Machine Learning in the science that relies on algorithms learning from data and solving tasks with minimal human intervention. Nowadays, as more and more information becomes available, machine learning algorithms need to move from traditional techniques to more scalable approaches that can use continuous stream of data. Complex machine learning systems usually can be decomposed into distinct modules that can be realized separately. A flat or hierarchical decomposition may be used to express such modularity. This thesis provides several modular online learning algorithms and their theoretical guarantees for different machine learning approaches that address the computational and statistical challenges. More specifically, we consider three machine learning categories: i) extreme multi-label classification, a supervised setting where the number of labels are extremely large: we propose algorithms that deploy hierarchical structure in the data or label space; ii) contextual bandits, a one-step reinforcement learning scenario, where the action space is continuous and infinite: we devise algorithms using hierarchical modularity in the action space; iii) task-agnostic continual learning in deep neural networks dealing with non-stationary data distribution imposed by sequence of distinct tasks: we introduce an online modular algorithm, where each module is responsible for an individual task. In the first part of this thesis, we consider extreme multi-label classification, where the goal is to annotate each data point with the most relevant subset of labels from an extremely large label set. Efficient annotation can be achieved with balanced hierarchical predictors, i.e. trees with logarithmic-depth in the label complexity, whose leaves correspond to labels. Designing prediction mechanism with such trees for real data applications is non-trivial as it needs to accommodate sending examples to multiple leaves while at the same time sustain high prediction accuracy. In this st