602 research outputs found
OPICO: A platform for collecting and analyzing emoji usage
No full textEmojis have emerged as a popular form of digital communication that can quickly convey big ideas or expressions with a single character. Since their creation in 1999, emojis have become a mainstay in the digital lexicography of instant messaging and social media. However, emojis are traditionally used in conjunction with text, as supplemental metadata. This paper explores the feasibility of emoji-only communication and emoji grammars. To gather emoji-only data, we built Opico, an emoji-first, social media mobile app that allows users to share emoji reactions about the places they visit with their friends. Each post in the app is a one-to-five emoji sentence to express an emotion, sentiment, or description about a location. After collecting over 3500 emoji reactions and having over 900 users on the app, the results demonstrate emoji-only communication provided concise forms of expression about various locations across the world. Data analysis has uncovered the various patterns and linguistics of how users construct emoji-only sentences, building the foundation for an emoji grammar.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2018-09-27 without embargo termsThe student, Sujay Khandekar, accepted the attached license on 2018-07-09 at 19:06.The student, Sujay Khandekar, submitted this Thesis for approval on 2018-07-09 at 19:13.This Thesis was approved for publication on 2018-07-10 at 12:14.DSpace SAF Submission Ingestion Package generated from Vireo submission #12786 on 2018-09-27 at 10:47:29Made available in DSpace on 2018-09-27T16:17:45Z (GMT). No. of bitstreams: 2
KHANDEKAR-THESIS-2018.pdf: 6939438 bytes, checksum: e5a686d2ff486f3c3a04f02c02ce84af (MD5)
LICENSE.txt: 4212 bytes, checksum: f4e88d0a9ff5771d56b5d402b28363fc (MD5)
Previous issue date: 2018-07-1
ResearchAgent: Iterative Research Idea Generation over Scientific Literature with Large Language Models
No full text
CCDI Toolkit: Diversity & Inclusion Councils
No full textDiversity and inclusion is a core leadership competency in today’s organizations. As an inclusive leader, I understand the need and value of diversity of thought. It is well documented that diversity of thought is vital to an organization’s operational success. However, success will not be achieved by diversity alone. Once you have diverse people in the organization, how do you create an inclusive culture? As leaders, we must look at how we can be inclusive to make sure that the benefits of having a diverse workforce contribute to the business success of our organizations. The Global Diversity and Inclusion Benchmark recommends executive-led diversity councils as a foundational structure for an inclusive organization. We are pleased to present the latest in our toolkit series Diversity and Inclusion Councils: Toolkit, which provides insight to having a properly structured and empowered diversity and inclusion council. In this toolkit, the author Sujay Vardhmane discusses two key pillars needed to create inclusive environments: 1. leaders who are committed to diversity and inclusion, and 2. the structures for successful diversity and inclusion councils. This toolkit defines diversity councils; describes the types; explains the value of diversity and inclusion councils to different areas of the organization and provides guidance on operationalizing diversity councils in your organization. It includes references to tools that will help you measure and report the results that will help your organization move ahead of its competition. The biggest takeaway for you the reader is the checklist for a successful diversity and inclusion council. Overall, this toolkit provides a framework that will help you implement a diversity and inclusion council to produce organizational results from an inclusive culture. We hope you enjoy and find value in this toolkit. We look forward to bringing you more tools and resources as we engage dedicated professionals across Canada to solve our biggest inclusion challenges. Thanks. Michael Bach, CCDP/AP Founder and CEO Canadian Centre for Diversity and Inclusio
Switch-on-to-fault scheme for transmission line protection
No full textSwitch-on-to-fault (SOTF) schemes are used to maintain dependability and speed when closing a transmission line breaker onto a faulted line. This is accomplished by enabling overreaching directional and nondirectional protection elements for a short window of time shortly after the transmission line breaker closes. When line potential transformers (PTs) are used to polarize directional distance relays, there is no benefit to using memory voltage during an SOTF condition and the polarizing signal used is directly related to the amount of fault voltage available. Depending on the magnitude of fault voltage available, the speed of a directional distance element can be quite slow, even for faults away from the PT location. To mitigate this dependability and speed issue, a nondirectional instantaneous overcurrent (50) element is typically used, sometimes with undervoltage (27) supervision to balance security.
This thesis uses a case study to illustrate the speed sacrifices made when a directional distance element, rather than an instantaneous overcurrent element, must trip during an SOTF condition. Results are provided from testing various directional distance elements to determine the minimum voltage required for fast operation of these elements. This information, is used to determine the lowest value to use for undervoltage supervision of the instantaneous overcurrent element to ensure a voltage-supervised 50 element has adequate reach for fast SOTF operation.
The benefits of using a nondirectional distance element for SOTF protection are discussed. This element is significantly easier to set than an undervoltage-supervised instantaneous overcurrent element, which helps to maintain dependability, security, and speed during SOTF conditions. To illustrate, we provide formulas to plot the reach of the 50 and 27 elements in the impedance plane so that we can directly compare to the nondirectional characteristic. A guidance formula on how to set 50 element to maintain dependability under single-contingency conditions is provided. For the 27 element, further setting optimization to maintain security during tapped loads and line charging current scenarios is illustrated. Additional considerations, including the SOTF duration timer window, security concerns for a sensitively set ground overcurrent element, resetting SOTF with healthy line voltage, SOTF benefits during the use of bus PTs, and high-speed reclosing are also discussed.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2022-08-01The student, Sujay Dasgupta, accepted the attached license on 2020-07-24 at 09:38.The student, Sujay Dasgupta, submitted this Thesis for approval on 2020-07-24 at 10:21.This Thesis was approved for publication on 2020-07-24 at 11:12.DSpace SAF Submission Ingestion Package generated from Vireo submission #15742 on 2020-10-02 at 15:34:11Made available in DSpace on 2020-10-07T22:44:48Z (GMT). No. of bitstreams: 2
DASGUPTA-THESIS-2020.pdf: 2127532 bytes, checksum: 343d65baad250e11bdb4e1d1c522cad0 (MD5)
LICENSE.txt: 4211 bytes, checksum: 45e5f6106d888f077040a09e91c7ddc4 (MD5)
Previous issue date: 2020-07-24Embargo set by: Seth Robbins for item 116269
Lift date: 2022-10-07T22:44:53Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Onl
PrivacyAlert: a dataset for image privacy prediction
No full textThis is a dataset for image privacy prediction. Images are from Flickr and annotated as private/public by crowd-sourcing platforms. Private images are ones that contain sensitive information and cannot be shared with everyone on social networking sites. Public images are ones that are safe to be shared with everyone. Our dataset can be used to train machine learning/deep learning models as binary classifiers to predict whether images contain sensitive information.
Please cite:
@inproceedings{zhao2022privacyalert,
title={PrivacyAlert: A Dataset for Image Privacy Prediction},
author={Zhao, Chenye and Mangat, Jasmine and Koujalgi, Sujay and Squicciarini, Anna and Caragea, Cornelia},
booktitle={Proceedings of the International AAAI Conference on Web and Social Media},
volume={16},
pages={1352--1361},
year={2022}
High resolution profiling of EGFR mutations in glioblastoma patients using an ultrasensitive digital PCR approach
No full textGlioblastoma Multiforme (GBM) is the most aggressive type of adult brain cancer. The average survival time after GBM diagnosis is 14.6 months even with current tri-modality therapy. The Epidermal Growth Factor Receptor (EGFR) is amplified in 57% of GBM. Mutations in EGFR such as EGFR variant III, A289V, and R108K lead to more aggressive tumors, and diminished survival. We are in dire need of a molecular assay that rapidly profiles these alterations in EGFR since other assays currently available clinically, like Next Generation Sequencing, may take up to 4 weeks due to the batching of samples in current workflows.
Our lab has established a very sensitive and novel digital Polymerase Chain Reaction (dPCR) assay that detects EGFRvIII in patient tumors within 24 hours of resection. This dPCR assay utilizes RNA extracted from microgram quantities of resected tumor from GBM patients, which is then converted to complementary DNA (cDNA). cDNA is then pre-amplified and subjected to the dPCR assay using specific primers and probes for EGFRvIII and EGFR WT. The assay is multiplexed with an internal reference control, RNaseP. The same starting material can be used to detect the presence or absence of two other mutations, R108K and A289V, with exquisite sensitivity and specificity.
We have utilized this assay and tested the platform on patient derived organoids and patient tumor samples. We have also validated this assay on exosomal RNA extracted from media used for culturing U87 WT and U87 vIII cell lines, as well as patient-derived glioma stem cell lines like NS039 and T4213.
This assay allows for rapid and ultrasensitive detection of EGFRvIII, EGFRWT, R108K, and A289V mutations in patient tumors and patient derived organoids. The workflow for this assay allows results within 24 hours of tumor resection, which facilitates early initiation of novel investigational therapeutic agents. It is possible that molecular characterization of tumor tissue, biofluids, microvesicles, platelets, and cfRNA would help to elucidate genomic variations that occur during disease recurrence. In the future, we plan to test this assay on RNA extracted from various microvesicles and platelets derived from blood to facilitate non-invasive tumor characterization and usefully complement conventional follow-up and imaging methods.This poster was presented at the first annual Celebration of Undergraduate Research and Creative Activity while the author was an undergraduate student at Rutgers University-Camden
Ultrasensitive molecular profiling of EGFR mutations in glioblastoma multiforme using a rapid & high-resolution digital PCR approach
No full textGlioblastoma Multiforme (GBM) is the most common and aggressive adult brain cancer with a 14.6-month average survival time, even with current therapies. 57% of GBM has amplified the Epidermal Growth Factor Receptor (EGFR). EGFR variant III, A289V, & R108K are mutations in EGFR that lead to increased tumor aggression and poorer prognosis. Learning more about these mutations will stop gliomas and prevent diminished survival, and shorter life expectancies in these patients. More quantitative methods of detection would allow for elucidation of mutations that occur during cancer development, with emphasis on metastasizing tumors. To inhibit EGFR in certain cancers, it is necessary to identify why increased proliferation occurs, usually because of mutation. After looking at the molecular changes in GBM, current methods of detection will be discussed, and finally, a novel method, digital Polymerase Chain Reaction (dPCR), will be introduced. While the detection of EGFR mutations in GBM tumors is arguably the best example for the utilization of dPCR, other applications of this technology will also be explored. Researchers have established very sensitive dPCR assays that detect various EGFR mutations in patient-derived tumors & organoids in only one day after resection surgery. RNA extracted from minute quantities of the patient tumor is converted to cDNA, which is then pre-amplified. Next, cDNA is run through a dPCR assay with a specially designed set of primers & probes for EGFR mutations and wild-type EGFR. In the future, RNA derived from microvesicles & platelets in a patient’s blood can be analyzed to help develop a less invasive characterization of tumors to eventually complement current diagnostic and treatment methods.Winner: First Place, 2021 Paul Robeson Library Undergraduate Research Award
Recommended from our members
Provable alternating minimization for non-convex learning problems
No full texttextAlternating minimization (AltMin) is a generic term for a widely popular approach in non-convex learning: often, it is possible to partition the variables into two (or more) sets, so that the problem is convex/tractable in one set if the other is held fixed (and vice versa). This allows for alternating between optimally updating one set of variables, and then the other. AltMin methods typically do not have associated global consistency guarantees; even though they are empirically observed to perform better than methods (e.g. based on convex optimization) that do have guarantees. In this thesis, we obtain rigorous performance guarantees for AltMin in three statistical learning settings: low rank matrix completion, phase retrieval and learning sparsely-used dictionaries. The overarching theme behind our results consists of two parts: (i) devising new initialization procedures (as opposed to doing so randomly, as is typical), and (ii) establishing exponential local convergence from this initialization. Our work shows that the pursuit of statistical guarantees can yield algorithmic improvements (initialization in our case) that perform better in practice.Electrical and Computer Engineerin
Data for figures in Kemp, E M, J W Wegiel, S V Kumar, J V Geiger, D M Mocko, J P Jacob, and C D Peters-Lidard, 2021: A NASA-Air Force precipitation analysis for near-real-time operations. Submitted to _J Hydrometeor_
No full text<p>Tar files containing gridded metrics, domain-wide metric means and confidence intervals, and rain-gauge reports used to generate figures in Kemp et al (2021).<br>
<br>
Citation:<br>
</p>
<p>Kemp, E M, J W Wegiel, S V Kumar, J V Geiger, D M Mocko, J P Jacob, and C D Peters-Lidard, 2021: A NASA-Air Force precipitation analysis for near-real-time operations. Submitted to _J Hydrometeor_.</p>
- …
