教学课程资源库 更多>>

摘要： Here are the grammatical subjects for each of these five sentences: (1) "There" is a word that just waves at the wind, referring to no one nor no thing; (2) "A statement"; (3) "Little that is new"; (4) "The progress of our arms"-no people here, just arms; and (5) "No prediction." [...]when all of that is accomplished, then pay attention to how the rhythms of each of your sub units within a sentence are sending instructions to your readers about how to balance this unit with some other unit. To present conflict intellectually instead of emotionally, he relies once again on his techniques of balance and control, raising distinctions by a combination of (1) focused repetition and (2) a rhythmic, musical structure that we can see at work in the colometrics. First sentence: At virtually every moment, Lincoln is using rhetorical techniques that have great effect upon his audience, without the audience necessarily being aware of any of them.

关键词： Bioengineering   Histology   Computer science   Surgery   Public health   Biomedical engineering   Osteopathic medicine  

摘要： Joint disorders significantly affect quality of life and present unique challenges for tissue engineering. In the craniofacial space, and especially for the temporomandibular joint (TMJ), there is an unmet need for anatomically precise and mechanically robust cartilage and bone tissues to recapitulate native function. Current surgical reconstruction methods, whether using autologous or synthetic options, suffer from imprecision, comorbidities, complications, and frequently require subsequent operations. Furthermore, many craniofacial graft efforts have focused on improving bone without addressing cartilage, which is essential to proper TMJ function. Thus, there is a compelling need to engineer a human-sized, biologically and anatomically matched cartilage-bone TMJ replacement. This dissertation demonstrates the ability to generate such a graft with native-like properties in a human-sized large animal model by focusing on two aims: (i) establish methods to fabricate and culture anatomically specific, autologous cartilage-bone grafts (Aim 1), and (ii) show improvement of graft performance after six months implantation in vivo compared to previous methods, controls, and native tissue (Aim 2). Using Yucatan mini-pigs as a human-sized model, the ramus-condyle unit (RCU), a geometrically intricate portion of the mandible and primary load bearing section of the TMJ, was targeted for reconstruction. Scaffolds were created using computer tomography (CT) image-guided micromilling of decellularized bone matrix, then infused with autologous adipose-derived chondrogenic and osteogenic progenitors. These biological constructs were then cultured in vitro in a novel dual-perfusion bioreactor before in vivo implantation. Similar in vitro culture of representative constructs done in parallel demonstrated cell attachment and some differentiation. After six months implantation, the dual cartilage-bone RCU grafts maintained their predefined anatomical structure and regenerated full-thic

关键词： Computer science   Artificial intelligence  

摘要： Neural Machine Translation (NMT) is the current trend approach in Natural Language Processing (NLP) to solve the problem of automatically inferring the content of target language given the source language. The ability of NMT is to learn deep knowledge inside languages by deep learning approaches. However, prior works show that NMT has its own drawbacks in NLP and in some research problems of Software Engineering (SE). In this work, we provide a hypothesis that SE corpus has inherent characteristics that NMT will confront challenges compared to the state-of-the-art translation engine based on Statistical Machine Translation. We introduce a problem which is significant in SE and has characteristics that challenges the ability of NMT to learn correct sequences, called Prefix Mapping. We implement and optimize the original SMT and NMT to mitigate those challenges. By the evaluation, we show that SMT outperforms NMT for this research problem, which provides potential directions to optimize the current NMT engines for specific classes of parallel corpus. By achieving the accuracy from 65% to 90% for code tokens generation of 1000 Github code corpus, we show the potential of using MT for code completion at token level.

关键词： Computer science   Mathematics education   User interface   Software   User needs   Playgrounds   Graphs   Labeling   Approximation   Educational technology   Algorithms   Heuristic   Traveling salesman problem   Editing  

摘要： A system called Graph Playground has been created to enhance the teaching of introductory graph theory and graph algorithms presented in courses like CSC 340, Applied Combinatorics, and CSC 440, Design and Analysis of Algorithms, at the University of Rhode Island. The primary goals of the system are to allow visual creation and manipulation of graphs, and to provide a set of algorithms that can be applied to graphs with intermediate steps shown as visual decorations of the graph. The features and capabilities of the system are described, with demonstrations of the application of algorithms on example graphs. The design and implementation details of the system are also discussed, along with discussion of challenges encountered in the implementation and how they were solved. The system is designed to be easily extensible for new algorithms, and a breakdown of the steps needed to implement a new algorithm is presented.

关键词： Science history   Computer science  

摘要： Americans love their history, and they certainly love their historical architecture. Architecture has always operated in between reality and fiction; the original and the copy, a precedent and an antecedent. history and future. Architect, and architectural representation, would seem to tend towards representations of truth, representations of the time in which they were produced. Architects are engrossed by notions of truth; truth in material, truth in form, etc. But what happens when truth is relative? What if the meanings of symbols, and their associations with time and history, do not matter - in fact, it is not allowed to matter? What happens when the our digital mechanisms fold multiple readings of history into one, without a beginning or end? Architecture and urbanism today desires to recycle, reuse and regurgitate a canon of precedent and standard; a canon or a code, (or perhaps a catalogue) is often critically suspect, and are often simply facsimiles divorced from meaning. But why? Why do Americans in particular obsess over particular styles, fetishized aesthetics or mass produced material approximations? These tend to be small-scale architectural components and details (column covers, statuettes, faux marble, screen printed wood textures, etc.) but also encompass the borrowing of whole projects and building forms, and are usually tied to some specific moment in history: Imperial Roman, Hellenistic Greece, Hindu, Judeo-Christian, Colonial American. Architecture has a long legacy of this, where resonances of projects fluidly move through time, constantly reinterpreting itself; this is essentially the critical intent of Postmodern architecture, but done ironically. One might even correlate the often-taught axiom, “Architects don’t design buildings, we design things that look like buildings.” Even further, one could say Architects design things that look like buildings, which look like other buildings.” While Postmodern architecture is critically ironic in its

关键词： Computer science   Artificial intelligence  

关键词： Computer science   Artificial intelligence  

关键词： Computer science   Artificial intelligence  

摘要： One of the important goals of Artificial Intelligence (AI) is to mimic the ability of humans to leverage the knowledge or skill from previously learned tasks to quickly learn a new task. For example, humans can reapply the learned skill of balancing the bicycle for learning to ride a motorbike. In a similar context, the field of Natural Language Processing (NLP) has several tasks including machine translation, textual summarization, image/video captioning, sentiment analysis, dialog systems, natural language inference, question answering, etc. While these different NLP tasks are often trained separately, leveraging the knowledge or skill from related tasks via joint training or training one task after another task in a sequential fashion, can have potential advantages. To this end, this dissertation explores various NLP tasks (especially multi-modal text generation and pair-wise classification tasks covering both natural language generation (NLG) and natural language understanding (NLU)) leveraging information from the related auxiliary tasks in an effective way via novel multi-objective learning strategies. These proposed novel learning strategies can be broadly classified into three paradigms: multi-task learning, multi-reward reinforcement learning, and continual learning. In multi-task learning, we mainly focus on intuitively finding what related auxiliary tasks can benefit the multi-modal video caption generation task and textual summarization task. We explore effective ways of sharing the parameters across these related tasks via joint training. In multi-reward reinforcement learning, we teach various skills to multi-modal text generation models in the form of rewards. For example, we try to teach the entailment skill to the video captioning model with entailment rewards. Further, we propose novel and effective ways of inducing multiple skills by `dynamically' choosing the auxiliary tasks (in MTL) or rewards (in RL) during the training in an automatic way usin

关键词： Bioinformatics   Biology   Computer science  

摘要： Plant phenotype descriptions are abundant both in literature and in community datastores. Enabling basic aggregation, organization, and analyses over this data requires that phenotype descriptions be represented in a computable format. One successful approach to this challenge has been to develop standardized vocabularies and biological ontologies that can be used to annotate phenotypes, allowing for the sparsity of the data to be reduced by inferring implicit information about the annotated data, and enabling simple quantification of similarity between annotated data. This type of structured curation has shown promise for enabling dataset-wide analyses on plant phenotype descriptions, but the time and effort required for curation of individual phenotype descriptions is a limiting factor in how scalable this approach is in light of the increasing volume of available text data related to plant phenotypes. Computational approaches have the potential to alleviate this problem by providing methods for representing phenotype descriptions and allowing quantification of phenotype similarity. In this work, computational pipelines for representing and comparing phenotypes are presented, and evaluated for their ability to predict biological relationships between genes. Approaches from the natural language processing domain perform as well as similarity metrics over curated annotations for predicting shared phenotypes. These approaches also show promise both for helping curators organize large datasets as well as for enabling researchers to explore relationships among available phenotype descriptions. A web application for querying datasets of plant phenotype descriptions and identifying associated genes is also presented, and example use cases are discussed.

关键词： Computer science   Design   Biostatistics  

摘要： Soft physical systems, be they elastic bodies, fluids, and compliant-bodied creatures, are ubiquitous in nature. Modeling and simulation of these systems with computer algorithms enable the creation of visually appealing animations, automated fabrication paradigms, and novel user interfaces and control mechanics to assist designers and engineers to develop new soft machines. This thesis develops computational methods to address the challenges emerged during the automation of the design, modeling, and control workflow supporting various soft dynamic systems. On the design/control side, we present a sketch-based design interface to enable non-expert users to design soft multicopters. Our system is endorsed by a data-driven algorithm to generate system identification and control policies given a novel shape prototype and rotor configurations. We show that our interactive system can automate the workflow of different soft multicopters' design, simulation, and control with human designers involved in the loop. On the modeling side, we study the physical behaviors of fluidic systems from a local, collective perspective. We develop a prior-embedded graph network to uncover the local constraint relations underpinning a collective dynamic system such as particle fluid. We also proposed a simulation algorithm to model vortex dynamics with locally interacting Lagrangian elements. We demonstrate the efficacy of the two systems by learning, simulating and visualizing complicated dynamics of incompressible fluid.