教学课程资源库 更多>>

关键词： Operations research   Health care management   Oncology   Computer science   Nursing  

摘要： Owing to treatment complexity in chemotherapy administration, nurses are usually required at the beginning, end, and at certain times during treatment to ensure high-quality infusion. It is, thus, critical for an outpatient chemotherapy unit to design a scheduling template that can effectively match nursing resources with treatment requirements. The template contains appointment slots of different lengths and thus allows schedulers to place patients into these appointment slots according to the provider’s order. As the template is often used over a period of several months, there usually exists a mismatch between the daily patient mix and the fixed structure of the given template. Hence, override policies must be employed to adjust to demand. However, these policies are often manually performed by schedulers. A Mixed-Integer Linear Programming (MILP) model has thus been proposed in this thesis to systematically develop optimal override policies in place of the manual process to improve template utilization while maintaining template stability. Numerical experiments based on real-life data from a chemotherapy unit are conducted to demonstrate the effectiveness of the proposed approach.

关键词： Therapy   Public health   Nuclear physics   Computer science  

摘要： The number of proton therapy facilities has been steadily increased since 1946, when Robert Wilson first introduced proton therapy. The advantageous properties of a proton beam for radiation therapy have generally been accepted for its ability to very precisely place dose in a tumor volume due to the characteristic of Bragg peak. Scattering systems and pencil beam scanning systems are two main proton beam delivery techniques used for proton therapy today. The MEVION S250i with HYPERSCAN and Adaptive Aperture (AA) pencil beam scanning (PBS) proton therapy system is the world’s first single-room gantry mounted active scanning proton therapy system, which delivers a pulsed beam with a nominal energy of 230 MeV recently installed and commissioned in our institution. The goal of radiation therapy is to deliver a sufficient dose to the tumor while limiting the dose to surrounding normal tissue. External beam radiation, such as photon or proton beam, is used for cancer treatment. The photon beams from a linear accelerator (LINAC) are routinely used for radiation therapy, such as intensity modulated radiation therapy (IMRT) and volumetric-modulated arc therapy (VMAT). Before we move on to the new proton therapy technique, we need to learn about designing and evaluating a treatment plan for photon therapy first. Our first goal has been to perform dosimetric comparisons between IMRT and VMAT for two different sites. For proton therapy, high energy protons travel through and interact with the beamline components. The radioactive materials are produced and can endanger the radiation safety of patients and the public. The smaller footprint of compact proton therapy systems such as the Mevion S250i proton therapy system has complications in the radiation protection consideration concept. To the best of our knowledge, there is no such report of its kind to address residual radioactivity from a clinical-operated particle therapy system, especially for a compact proton therapy syste

摘要： Personalized learning is gaining popularity in online computer science education due to its characteristics of pacing the learning progress and adapting the instructional approach to each individual learner from a diverse background. Among various instructional methods in computer science education, hands-on labs have unique requirements of understanding learners' behavior and assessing learners' performance for personalization. Hands-on labs are a critical learning approach for cybersecurity education. It provides real-world complex problem scenarios and helps learners develop a deeper understanding of knowledge and concepts while solving real-world problems. But there are unique challenges when using hands-on labs for cybersecurity education. Existing hands-on lab exercises materials are usually managed in a problem-centric fashion, while it lacks a coherent way to manage existing labs and provide productive lab exercising plans for cybersecurity learners. To solve these challenges, a personalized learning platform called ThoTh Lab specifically designed for computer science hands-on labs in a cloud environment is established. ThoTh Lab can identify the learning style from student activities and adapt learning material accordingly. With the awareness of student learning styles, instructors are able to use techniques more suitable for the specific student, and hence, improve the speed and quality of the learning process. ThoTh Lab also provides student performance prediction, which allows the instructors to change the learning progress and take other measurements to help the students timely. A knowledge graph in the cybersecurity domain is also constructed using Natural language processing (NLP) technologies including word embedding and hyperlink-based concept mining. This knowledge graph is then utilized during the regular learning process to build a personalized lab recommendation system by suggesting relevant labs based on students' past learning history to maxim

关键词： Aerospace engineering   Artificial intelligence   Computer science   Electrical engineering   Engineering   Physics  

摘要： This thesis focuses on sparse array architectures for the next generation of wireless communication, known as fifth-generation (5G), and automotive radar direction-of-arrival (DOA) estimation. For both applications, array spatial resolution plays a critical role to better distinguish multiple users/sources. Two novel base station antenna (BSA) configurations and a new sparse MIMO radar, which both outperform their conventional counterparts, are proposed. We first develop a multi-user (MU) multiple-input multiple-output (MIMO) simulation platform which incorporates both antenna and channel effects based on standard network theory. The combined transmitter-channel-receiver is modeled by cascading Z-matrices to interrelate the port voltages/currents to one another in the linear network model. The herein formulated channel matrix includes physical antenna and channel effects and thus enables us to compute the actual port powers. This is in contrast with the assumptions of isotropic radiators without mutual coupling effects which are commonly being used in the Wireless Community. Since it is observed in our model that the sum-rate of a MU-MIMO system can be adversely affected by antenna gain pattern variations, a novel BSA configuration is proposed by combining field-of-view (FOV) sectorization, array panelization and array sparsification. A multi-panel BSA, equipped with sparse arrays in each panel, is presented with the aim of reducing the implementation complexities and maintaining or even improving the sum-rate. We also propose a capacity-driven array synthesis in the presence of mutual coupling for a MU-MIMO system. We show that the appearance of grating lobes is degrading the system capacity and cannot be disregarded in a MU communication, where space division multiple access (SDMA) is applied. With the aid of sparsity and aperiodicity, the adverse effects of grating lobes and mutual coupling are suppressed and capacity is enhanced. This is performed by proposing a

关键词： Electrical engineering   Computer engineering   Biomedical engineering   Neurosciences   Computer science   Bioinformatics  

摘要： Vibrotactile sleeves and multimodal armbands show promise as devices that can transmit information to a user through the tactile sense. In this way, individuals have the potential to receive information haptically when typical auditory or visual channels are preoccupied or unavailable. To achieve this, individuals must successfully learn the mapping between haptic cues and informational icons through cross-modal associative learning. The success of this process is limited by perceptual capabilities of users, as well as lack of neural markers to quantify the success of haptic learning. In order to optimize future wearable displays and training methods necessary to maximize the potential of a haptic communication channel via the arm, we need to better understand how to improve perceptibility of tactile cues transmitted to users, as well as develop neural correlates to track haptic learning. This thesis contributes to these goals in two ways. First, we evaluate perceptual capabilities of individuals using a vibrotactile sleeve, investigating how the parameter of cue duration may affect localizability of cues presented along the forearm. Second, we evaluate a Representational Similarity Analysis approach as a way to track neural correlates of haptic learning on a multimodal arm display. Results will be used to inform the design of tactile cues and training paradigms for cross-modal associative learning applications.

摘要： From Counter-Memory to Counter-Culture: Black Islam in the U.S. Through a Mundane Afrofuturist Lens is a multi-generational look at moments of Mundane Afrofuturism within the lives of Black Muslims in the U.S., emphasizing the concepts of counter-memory and counter-culture as key cultural and political approaches toward subversion and innovation. Consisting of three artworks, the exhibition space is intended to act as an unveiling of information, recognizing Black folks’ historical and ongoing relationship to accessing knowledge.

关键词： Computer science   Computer engineering   Information technology   Artificial intelligence  

摘要： The aim of this dissertation is to develop constraint-based methods that extend and improve on current deep learning neural networks such as transformers and sequence-to-sequence (seq2seq) models, for the problem of question generation based on the analysis of the text of legal agreements, particularly privacy policies. A privacy policy is a legally binding agreement between a customer and service provider. This dissertation focuses on analyzing a privacy policy document to generate questions that capture entities and the relationships between them. Another area of focus is the generation of constraints based on domain knowledge and their application to the deep learning network during the question generation process. A possible use case of this research is development of test corpus for question answering systems in the privacy domain because the shortage of sufficiently large corpora poses a key challenge in the development of question answering and question generation systems. Question generation is the task of generating an interrogative sentence based on some text. Current approaches to question generation use sequence-to-sequence models with additional information like answers, positions of the answers, part-of-speech details, named entity tags among others. The idea behind such approaches is that these models can benefit from additional information about the text (i.e., sentence or paragraph). Recently, transformer-based approaches that offer the benefit of attention mechanism have also been used for generating questions. Transformers have achieved state-of-the-art results in many natural language processing tasks including text classification, machine translation, language understanding, co-reference resolution, and summarization. However, the contribution of transformers towards a task like question generation has not been as significant. This research tries to find ways of improving existing approaches by injecting domain knowledge, modeled as a combinatio

关键词： Artificial intelligence   Computer science  

摘要： The unit selection problem aims to identify a set of individuals who are most likely to exhibit a desired mode of behavior, which is defined in counterfactual terms. A typical example is that of selecting individuals who would respond one way if encouraged and a different way if not encouraged. Unlike previous works on this problem, which rely on ad-hoc heuristics, we approach this problem formally, using counterfactual logic, to properly capture the nature of the desired behavior. This formalism enables us to derive an informative selection criterion which integrates experimental and observational data. We show that a more accurate selection criterion can be achieved when structural information is available in the form of a causal diagram. We further discuss data availability issue regarding the derivation of the selection criterion without the observational or experimental data. We demonstrate the superiority of this criterion over A/B-test-based approaches.

关键词： Design   Codes   Energy   Algorithms   Error correction & detection   Fault tolerance   Expected values   Optimization   Confidentiality   Benchmarks   Computer science  

摘要： Recent trends in supercomputing, shared cloud computing, and “big data” applications are placing ever-greater demands on computing systems and their memories. Such applications can readily saturate memory bandwidth, and often operate on a working set which exceeds the capacity of modern DRAM packages. Unfortunately, this demand is not matched by DRAM technology development. As Moore’s Law slows and Dennard Scaling stops, further density improvements in DRAM and the underlying semiconductor devices are difficult [1]. In anticipation of this limitation, researchers have pursued emerging memory technologies that promise higher density than conventional DRAM devices. One such technology in phase-change memory (PCM) is especially desirable due to its increased density relative to DRAM. However, this nascent memory has outstanding challenges to overcome before it is viable as a DRAM replacement. PCM devices have limited write endurance, and can consume more energy than their DRAM counterparts, necessitating careful control of how and how often they are written. A second challenge is the non-volatile nature of PCM devices; many applications rely on the volatility of DRAM to protect security critical applications and operating system address space between accesses and power cycles. An obvious solution is to encrypt the memory, but the effective randomization of data is at odds with techniques which reduce writes to the underlying memory. This body of work presents three contributions for addressing all challenges simultaneously under the assumption that encryption is required. Using an encryption and encoding technique called CASTLE & TOWERs, PCM can be employed as main memory with up to 30× improvement in device lifetime while opportunistically reducing dynamic energy. A second technique called MACE marries encoding and traditional error-correction schemes providing up to 2.6× improvement in device lifetime alongside a whole-lifetime energy evaluation framework to guide sy

关键词： Computer science   Artificial intelligence   Logic  

摘要： Symbolic knowledge representation and reasoning and deep learning are fundamentally different approaches to artificial intelligence with complementary capabilities. The former are transparentand data-efficient, but they are sensitive to noise and cannot be applied to non-symbolic domains where the data is ambiguous. The latter can learn complex tasks from examples, are robust to noise, but are black boxes; require large amounts of --not necessarily easily obtained-- data, and are slow to learn and prone to adversarial examples. Either paradigm excels at certain types of problems where the other paradigm performs poorly. In order to develop stronger AI systems, integrated neuro-symbolic systems that combine artificial neural networks and symbolic reasoning are being sought. In this context, one of the fundamental open problems is how to perform logic-based deductive reasoning over knowledge bases by means of trainable artificial neural networks. Over the course of this dissertation, we provide a brief summary of our recent efforts to bridge the neural and symbolic divide in the context of deep deductive reasoners. More specifically, We designed a novel way of conducting neuro-symbolic through pointing to the input elements. More importantly we showed that the proposed approach is generalizable across new domain and vocabulary demonstrating symbol-invariant zero-shot reasoning capability. Furthermore, We have demonstrated that a deep learning architecture based on memory networks and pre-embedding normalization is capable of learning how to perform deductive reason over previously unseen RDF KGs with high accuracy. We are applying these models on Resource Description Framework (RDF), first-order logic, and the description logic EL+ respectively. Throughout this dissertation we will discuss strengths and limitations of these models particularly in term of accuracy, scalability, transferability, and generalizabiliy. Based on our experimental results, pointer networks pe