6501 research outputs found
Sort by
Adversarial Robustness with Partial Isometry
Get PDFDespite their remarkable performance, deep learning models still lack robustness guarantees, particularly in the presence of adversarial examples. This significant vulnerability raises concerns about their trustworthiness and hinders their deployment in critical domains that require certified levels of robustness. In this paper, we introduce an information geometric framework to establish precise robustness criteria for (Formula presented.) white-box attacks in a multi-class classification setting. We endow the output space with the Fisher information metric and derive criteria on the input–output Jacobian to ensure robustness. We show that model robustness can be achieved by constraining the model to be partially isometric around the training points. We evaluate our approach using MNIST and CIFAR-10 datasets against adversarial attacks, revealing its substantial improvements over defensive distillation and Jacobian regularization for medium-sized perturbations and its superior robustness performance to adversarial training for large perturbations, all while maintaining the desired accuracy
Italian Instrumental Music as Fascist Propaganda in the United States during the Interwar Period
Get PDFReframing Wikipedia as an information source: Concepts and strategies for critical inquiry and digital literacy.
Get PDFSince its inception in 2001, Wikipedia has grown to become the largest and most popular reference source in the world. Contrary to the idea that it lives in the Internet’s Wild West, this resource has robust infrastructure, with editorial guidelines, processes, and dedicated volunteers who help to make Wikipedia an invaluable information source for learning about almost any topic. Of course, it’s also true that not every Wikipedia article is reliable, and that the quality and depth of articles ranges. The crowdsourced nature of Wikipedia, along with its extensive editorial processes, make it not only a powerful (albeit imperfect) reference source, but also a rich tool for teaching critical source evaluation and critical understandings of how online information is created, circulated, and negotiated. The presenter will highlight key concepts and strategies from their “Using Wikipedia Wisely” workshop that can be integrated into teaching critical inquiry and digital literacies across the curriculum
Blueberry Drone AI: Estimating Crop Yield using Deep Learning & Smart Drones
Get PDFThis project seeks to assist blueberry growers in New Jersey estimate crop yield by developing software that allows autonomous drones to capture aerial images of blueberry bushes in the field, perform berry count, and identify blueberry conditions using deep learning models & computer vision
Transit Access and Mobility Challenges Among Rowan University Students
Get PDFCollege students can face unique mobility challenges and are an essential group of riders who rely on public transportation. For commuting students, public transportation or the campus shuttle can be one of the most relied-on ways to get to school. For on-campus students, access to public transportation can play an important role in their well-being and academic performance. This project conducts a comprehensive survey study to understand the access to transit service, mobility challenges, and impacts on student well-being and academic performance at Rowan University
MissionLog R - Air Force Mission History Report Management System with Encryption & Database Integration
Get PDFMissionLog R is a mobile and web application designed to allow for the filing, confirmation, and finalization of a mission history report for the US Air Force. The mobile application is used by filing members to fill out their report on their issued iPad. The form has been fitted with data validation and ghost data so the user knows what type of input that specific field is looking for. The web application can be used by the filing member to import their report from their iPad if no WiFi was available when they filled out the report. Then, the confirming and finalizing members can reject or approve a given report. Both applications are linked to a database where the confirming and finalizing members have a queue of reports, sorted by time when the report was submitted
WaterCAVE: Virtual Reality for Engaging the Next Generation in Wastewater Management
Get PDFWithin the next five to ten years, a significant portion of water and wastewater personnel are expected to retire, posing a substantial challenge for the utilities industry. The potential shortage of experienced workers poses inherent risks to the continuity of water services. WaterCAVE emerges as one of several tools designed to address this impending issue by engaging and inspiring the next generation to consider careers in the wastewater industry. This virtual reality (VR) simulation utilizes a ten-sided immersive display and was designed using Unity\u27s 3D real-time development platform. Students from various educational levels, ranging from Pre-K to 12, are invited to participate in field trips at the Rowan Technology Park, where they can explore a virtual wastewater treatment plant. WaterCAVE introduces students to a customizable virtual tour guide, enabling them to navigate through the intricacies of wastewater treatment and interact with different personnel involved in operating a wastewater treatment facility. Through its innovative approach, WaterCAVE aims to spark curiosity and interest among the next generation, fostering a potential influx of future professionals into the wastewater industry
Combined effects of exercise and immuno-chemotherapy treatments on tumor growth in MC38 colorectal cancer-bearing mice.
Get PDFAcute exercise induces transient modifications in the tumor microenvironment and has been linked to reduced tumor growth along with increased infiltration of immune cells within the tumor in mouse models. In this study, we aimed to evaluate the impact of acute exercise before treatment administration on tumor growth in a mice model of MC38 colorectal cancer receiving an immune checkpoint inhibitor (ICI) and chemotherapy. Six-week-old mice injected with colorectal cancer cells (MC38) were randomized in 4 groups: control (CTRL), immuno-chemotherapy (TRT), exercise (EXE) and combined intervention (TRT/EXE). Both TRT and TRT-EXE received ICI: anti-PD1-1 (1 injection/week) and capecitabine + oxaliplatin (5 times a week) for 1 week (experimentation 1), 3 weeks (experimentation 2). TRT-EXE and EXE groups were submitted to 50 minutes of treadmill exercise before each treatment administration. Over the protocol duration, tumor size has been monitored daily. Tumor growth and microenvironment parameters were measured after the intervention on Day 7 (D7) and Day 16 (D16). From day 4 to day 7, tumor volumes decreased in the EXE/TRT group while remaining stable in the TRT group (p=0.0213). From day 7 until day 16 tumor volume decreased with no significant difference between TRT and TRT/EXE. At D7 the TRT/EXE group exhibited a higher total infiltrate T cell (p=0.0118) and CD8+ cytotoxic T cell (p=0.0031). At D16, tumor marker of apoptosis, vascular integrity and inflammation were not significantly different between TRT and TRT/EXE. Our main result was that acute exercise before immuno-chemotherapy administration significantly decreased early-phase tumor growth (D0 to D4). Additionally, exercise led to immune cell infiltration changes during the first week after exercise, while no significant molecular alterations in the tumor were observed 3 weeks after exercise
DYNAMICS OF PROFICIENCY IN HIGH-TECH SYMBOL-BASED AAC: A MULTIPLE CASE STUDY EXPLORING THE PROCESS OF FIVE STUDENTS BECOMING PROFICIENT AAC USERS WITH SUPPORT FROM FAMILY AND SCHOOL TEAMS
No full textIn this qualitative multiple case study, I collaborated with five students who identified as proficient high-tech symbol-based (HTSB) AAC users, along with seven of their school team members (STMs) and five of their (FTMs) to explore the process of learning to communicate proficiently using HTSB AAC. Data collection involved three rounds of accessible interviews with student AAC users, single interviews with their trusted STMs and FTMs, and an examination of artifacts and documents. Detailed case analyses were presented for each of the five cases in the study, which centered on students’ perspectives and experiences. The results from individual case analyses were extended to compare findings within participant groups and across cases. In this dissertation, I present the following findings: (a) students’ priorities as proficient HTSB AAC users, (b) key strategies and guiding tenets of STMs, (c) guiding principles of FTMs, (d) foundational strategies and guiding tenets shared by both STMs and FTMs, and (e) collective priorities of proficient HTSB AAC users, STMs, and FTMs. Lastly, I describe the process of students becoming proficient HTSB AAC users through three phases: (a) access, (b) learning, and (c) proficiency. The results of this study have notable implications for research, practice, policy, AAC software and hardware developers, and most importantly the AAC user community
Development of a novel humanized anti-TSLP monoclonal antibody HZ-1127 with anti-allergic diseases and cancer potential.
Get PDFThymic stromal lymphopoietin (TSLP) is a member of the IL-2 cytokine family and has been widely recognized as a master regulator of type 2 inflammatory responses at barrier surfaces. Recent studies found dysregulation of the TSLP-TSLP receptor (TSLPR) pathway is associated with the pathogenesis of not only allergic diseases but also a wide variety of cancers including both solid tumors and hematological tumors. Thus, the blockade of TSLP represents an attractive therapeutic strategy for allergic diseases and cancer. In this study, we report the development of a novel humanized anti-TSLP monoclonal antibody (mAb) HZ-1127. Binding affinity, specificity, and ability of HZ-1127 in inhibiting TSLP were tested. HZ-1127 selectively binds to the TSLP cytokine with high affinity and specificity. Furthermore, HZ-1127 dramatically inhibits TSLP-dependent STAT5 activation and is more potent than Tezepelumab, which is an FDA-approved humanized mAb against TSLP for severe asthma treatment in inhibiting TSLP-induced CCL17 and CCL22 chemokines secretion in human peripheral blood mononuclear cells. Our pre-clinical study demonstrates that HZ-1127 may serve as a potential therapeutic agent for allergic diseases and cancer