skip to main content

Search for: All records

Creators/Authors contains: "Joshi, A."

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Free, publicly-accessible full text available January 1, 2023
  2. Scholarly literature on the concept of entrepreneurial ecosystems has increased sharply over the past five years. The surge in interest has also heightened the demand for robust empirical measures that capture the complexity of dynamic relationships among ecosystem constituents. We offer a framework for measurement that places collaborative relationships among entrepreneurs, firms, government agencies, and research institutions at the center of the ecosystem concept. We further emphasize the four roles of the federal government as a catalyst, coordinator, certifier, and customer in shaping these relationships. Despite the central importance of these firm-government interactions, there is surprisingly little research on suitablemore »methodologies and appropriate data for systematically and reliably incorporating them into measures of ecosystem health. Our study aims to address this gap in the literature by first developing a conceptual framework for measuring entrepreneurial ecosystems and then describing an array of accompanying databases that provide rich and detailed information on firms and their relationships with government organizations, accelerators, and research institutions. A major advantage of our approach is that all the underlying databases are drawn from non-confidential, publicly available sources that are transparently disclosed and regularly updated. This greatly expands the potential community of scholars, managers, and policymakers that may independently use these databases to test theories, make decisions, and formulate policies related to innovation and entrepreneurship.« less
    Free, publicly-accessible full text available September 29, 2022
  3. Free, publicly-accessible full text available December 1, 2022
  4. The paradigm of differentiable programming has significantly enhanced the scope of machine learning via the judicious use of gradient-based optimization. However, standard differentiable programming methods (such as autodiff) typically require that the machine learning models be differentiable, limiting their applicability. Our goal in this paper is to use a new, principled approach to extend gradient-based optimization to functions well modeled by splines, which encompass a large family of piecewise polynomial models. We derive the form of the (weak) Jacobian of such functions and show that it exhibits a block-sparse structure that can be computed implicitly and efficiently. Overall, we showmore »that leveraging this redesigned Jacobian in the form of a differentiable" layer''in predictive models leads to improved performance in diverse applications such as image segmentation, 3D point cloud reconstruction, and finite element analysis. We also open-source the code at\url {https://github. com/idealab-isu/DSA}.« less
    Free, publicly-accessible full text available December 1, 2022
  5. Cyber-defense systems are being developed to automatically ingest Cyber Threat Intelligence (CTI) that contains semi-structured data and/or text to populate knowledge graphs. A potential risk is that fake CTI can be generated and spread through Open-Source Intelligence (OSINT) communities or on the Web to effect a data poisoning attack on these systems. Adversaries can use fake CTI examples as training input to subvert cyber defense systems, forcing the model to learn incorrect inputs to serve their malicious needs. In this paper, we automatically generate fake CTI text descriptions using transformers. We show that given an initial prompt sentence, a publicmore »language model like GPT-2 with fine-tuning, can generate plausible CTI text with the ability of corrupting cyber-defense systems. We utilize the generated fake CTI text to perform a data poisoning attack on a Cybersecurity Knowledge Graph (CKG) and a cybersecurity corpus. The poisoning attack introduced adverse impacts such as returning incorrect reasoning outputs, representation poisoning, and corruption of other dependent AI-based cyber defense systems. We evaluate with traditional approaches and conduct a human evaluation study with cybersecurity professionals and threat hunters. Based on the study, professional threat hunters were equally likely to consider our fake generated CTI as true.« less
  6. With computing impacting most every professional field, it has become essential to provide pathways for students other than those majoring in computer science to acquire computing knowledge and skills. Virtually all employers and graduate and professional schools seek these skills in their employees or students, regardless of discipline. Academia currently leans towards approaches such as double majors or combined majors between computer science and other non-CS disciplines, commonly referred to as “CS+X” programs. These programs tend to require rigorous courses gleaned from the institutions’ courses for computer science majors. Thus, they may not meet the needs of majors in disciplinesmore »such as the social and biological sciences, humanities, and others. The University of Maryland, Baltimore County (UMBC) is taking an approach more suitably termed “X+CS” to fulfill the computing needs of non-CS majors. As part of a National Science Foundation (NSF) grant, we are developing a “computing” minor specifically to meet their needs. To date, we have piloted the first two of the minor’s approximately six courses. The first is a variation on the existing Computer Science I course required for majors but restricted to nonmajors. Both versions of the course use the Python language and cover the same programming content, but with the non-majors assigned projects with relevance to non-CS disciplines. We use the same student assessment measures of homework, projects, and examinations for both courses. After four semesters, results show that non-CS majors perform comparably to majors. Students also express increased interest in computing and satisfaction with being part of a non- CS major cohort. The second course was piloted in fall 2019. It is a new course intended to enhance and hone programming skills and introduce topics such as web scraping, HTML and CSS, web application development, data formats, and database use. Students again express increased interest in computing and were already beginning to apply the computing skills that they were learning to their non-CS courses. As a welcome side effect, we experienced a significant increase in the number of women and under-represented minorities (URMs) in these two courses when compared with CS-major specific courses. Overall, women comprised 52% of the population, with URMs following a similar upward trend. We are currently developing the third course in the computing minor and exploring options for the remaining three. Possibilities include electives from our Information Systems major. We will also be working with our science, social science, and humanities departments to utilize existing courses in those disciplines that apply computing. The student response that we have received thus far provides us with evidence that our computing minor will be popular among UMBC’s non-CS population, providing them with a more suitable and positive computing education than existing CS+X efforts.« less