1854 research outputs found
Sort by
Review of text classification methods on deep learning
No full textText classification has always been an increasingly crucial topic in natural language processing. Traditional text classification methods based on machine learning have many disadvantages such as dimension explosion, data sparsity, limited generalization ability and so on. Based on deep learning text classification, this paper presents an extensive study on the text classification models including Convolutional Neural Network-Based (CNN-Based), Recurrent Neural Network-Based (RNN-based), Attention Mechanisms-Based and so on. Many studies have proved that text classification methods based on deep learning outperform the traditional methods when processing large-scale and complex datasets. The main reasons are text classification methods based on deep learning can avoid cumbersome feature extraction process and have higher prediction accuracy for a large set of unstructured data. In this paper, we also summarize the shortcomings of traditional text classification methods and introduce the text classification process based on deep learning including text preprocessing, distributed representation of text, text classification model construction based on deep learning and performance evaluation
Blockchain-Based Secure and Privacy-Preserving Clinical Data Sharing and Integration
No full textThis paper is an exploration to securely integrate geo-scattered medical data silos and to provide privacy-preserving data sharing via blockchain and cryptography. We leverage broadcast encryption, key regression, data privacy classification, blockchain and smart contract holistically to address such a challenge. In our design, a patient’s medical records are divided into multiple parts with different sensitivities according to personal privacy requirements. Each part is encrypted with a symmetric encryption key and it is broadcast encrypted to a pre-defined user set. Data encryption keys are generated by a key regression scheme, where keys used to encrypt low-sensitivity data parts can be derived from keys used to encrypt high-sensitivity data parts. With such a mechanism, we can enforce strict access and privacy control on medical data. Furthermore, we use Ethereum blockchain to connect various institutions to provide efficient data sharing, and we design smart contracts to implement the business logic of medical data sharing
Institutions, entrepreneurial adaptation, and the legal form of the organization
No full textPurpose: This paper aims to examine how economic freedom impacts the initial choice of legal structure for startup firms. The authors do this by first exploring whether economic freedom is an essential determinant of the initial legal form of organization (LFO). The authors then explore the impact of economic freedom on firms\u27 choice of changing their initial legal structure over time and how this change impacts their survival rate. Design/methodology/approach: The authors employ a multinomial logistic regression model to measure the initial determinants of LFO by utilizing an eight-year panel data set of 4,928 startups in the USA through the Kauffman firm survey and merge it with the Economic Freedom in North American index from the Fraser Institute. The authors then employ a logistic regression model to examine the determinants facilitating a change in legal structure over time. Findings: The results show that economic freedom is a significant determinant in the choice of legal structure. The findings also report that the majority of startups do not change their legal form, but of those that do change the legal structure show a higher survival rate. Research limitations/implications: Major limitations are the size of the data and the nature of somewhat limited economic freedom differences with the USA. More nuanced measures of economic freedom would be highly desirable. Practical implications: Policymakers should take note that limited red tape, smoothly working labor markets and straightforward processes for changes of legal structures of organizations would improve survival and growth odds for entrepreneurs. Originality/value: Drawing on the theory of institutions, the authors attempt to bridge a gap in the literature by explicitly analyzing the determinants of the legal structure in startups in light of economic freedom. Institutional factors do not work in isolation; therefore, the authors also employ traditional entrepreneur-specific variables that affect the choice of legal structure in addition to the institutional framework
Upcycling Plastic Via Thermal Compression
Get PDFEach year Americans throw away 35 billion plastic bottles. The goal of our project is to demonstrate an alternative to simply discarding unwanted plastic. Our process shows how plastic bottles can be transformed into coasters by using a blender, mold, and heat press. We focused on high-density polyethylene (HDPE) since it is readily available and safe to work with. HDPE is often thrown away because it is not always economical to recycle due to its low yield. Many hours of testing and visits to the fabrication lab taught us about the tribulations of upcycling plastic and the complexities that come with creating a mold for new products. We conducted multiple blending, heating, cooling, pressing, and extracting tests to determine the optimal plastic piece size, press conditions, and mold transferring methods. Fall semester, we focused on designing a process for collecting and processing plastic while determining ideal operating conditions for the heat press. This spring, we altered old molds, designed new molds, and tested rapid cooling methods. Our ability to design a process that produces one coaster in approximately 30 minutes using only a heat-press proves that this alternative solution to plastic waste has great potential to be scaled up
Chemotherapeutic sensitivity and the expression of BIN1 in Human Diploid Fibroblasts expressing SV40 T-antigen
Get PDFCisplatin is the chemotherapeutic of choice to treat several cancers, notably cervical and esophageal, which are commonly induced by the tumor virus HPV. Cisplatin’s mechanism of action is to create intra-strand and inter-strand linkages between DNA; however some tumors have been shown to have developed resistance. It has been postulated that resistance to cisplatin can be linked to BIN1 expression levels or the isoforms created as the result of splicing by SRSF1. BIN1 (bridging integrator-1) is involved in many functions of the cell, but most importantly it regulates the cell cycle and apoptosis through its tumor suppressor characteristics. To help elucidate the mechanisms involved in cancer induction by tumor viruses and help further chemotherapeutic treatment options, this research was undertaken to determine if cisplatin was effective against cells transformed with the Simian Virus 40 (SV40) T antigen and to determine if the mechanism of action is linked to BIN1 expression levels. To accomplish this, human diploid fibroblast immortalize with telomerase [HDF(tert)] were transfected with a plasmid encoding the SV40 T antigen; and, two distinct clones were expanded [HDF(tert)+T Clone 1 and HDF(tert)+T Clone 2]. All cells were exposed to cisplatin and growth was assessed. It was found that cells expressing the oncoprotein were more sensitive to cisplatin than the parental cell line. To determine if the parental line resistance to cisplatin correlated to BIN1, expression level and isoform production was assessed. This research found that there is no difference in the accumulated levels of BIN1 protein being expressed among the 3 cell lines as determined by Western blotting. However, RT-PCR revealed multiple isoforms being expressed. The predominant form in all cell lines appeared to be isoform 10 which correlates to the 456 base pair band and the 45 kDa protein band. Interestingly, a new, ninth unpublished isoform was isolated which was present in both the immortalized parental line [HDF(tert)] and the transfected cell lines [HDF(tert)+T Clone 1 and HDF(tert)+T Clone 2]. Overall this data indicated that BIN1 expression is not solely responsible for cisplatin effectiveness. T-antigen-oncoprotein-expressing-cells were killed up to 90%; yet, BIN1 levels were equal to that of the non-transformed line. Whereas the non-transformed cells exhibited 50% survival. It is important to determine the mechanism for cisplatin sensitivity in order to apply it to make cancer cells less resistant to treatment
Changes in Time
No full textWe talk about different technologies and how they change over time. Near the end, we get to know each other better by talking about more personal and deeper questions
Flash Prose, June 8th.
Get PDFA flash prose piece I wrote for a writing contest during the Covid-19 Pandemic written at home (Orefield, PA) on June 8th, 2020
The Crows.
Get PDFThis poem was written on June 8th as a response to the confusing and momentous feelings accompanying this pandemic
Dispatches From Student Quarantine, episode 301
No full textSeries of 21 episodes created by COM220. Includes contributions by students Tea Ceresini, Kaitlyn Chambers, Jessica Freels, Sarah Hasenauer, Emily Kuhn, Rachel Little, Olivia Moyer, Patrick Osborn, Rebecca Parsons, Chad Rosenberger, Cameron Scandle, Samantha Seely, William Snyder, Christopher Tongel, and Kevin Wenger