Search for: All records

Engaging undergraduates in research has been shown to improve retention, increase students' sense of computer science identity, and increase their chances of continuing to graduate school. Yet research experiences at most universities are ad hoc, and many undergraduates-particularly those from groups underrepresented in computing-do not have the opportunity to participate. The Early Research Scholars Program (ERSP) is a structured, academic-year group-based undergraduate research program designed to help universities vastly increase participation in research for early computing undergraduates. ERSP launched at UC San Diego in 2014 where it now annually engages over 50 second-year undergraduates, 59% of whom are women, and 22% of whom are from underrepresented racial and ethnic groups. The program's portable design has enabled its expansion to 7 other colleges and universities. This workshop will train participants in launching ERSP (or any part of it) at their university to increase and diversify the undergraduates participating in research. Workshop leaders are the ERSP directors at four universities. They will address how to launch and run the program in different contexts. They will provide an interactive, hands-on experience of running the program covering the following topics: developing and teaching a research methods class, student application and selection to ensure a diverse and supportive cohort, and creating a dual-mentoring structure to engage and retain early undergraduates without overburdening faculty. Workshop participants will be invited to join the ERSP virtual community to get support launching their own version of ERSP. 

The Early Research Scholars Program (ERSP) was launched in 2014 at UC San Diego as a way to provide the benefits of research experiences to a large and diverse group of students early in their undergraduate computing career. ERSP is a structured program in which second-year undergraduate computing majors participate in a group-based, dual-mentored research apprenticeship over a full academic year. In its first four years ERSP engaged 139 students with a high proportion of women (68%) and racially minoritized students (19%), and participation in ERSP correlated with increased class grades. In 2018 we partnered with three additional universities to launch their own version of ERSP. Implementations at our partner sites have seen similar diversity and initial success, and have taught us how to implement the program in different contexts (e.g. quarters vs. semesters, different credit structures). This paper describes the structure of ERSP and how it can be adapted to different contexts to construct a scalable and inclusive research experience for early-career undergraduates in computing and related fields. 

Undergraduate research experiences are a promising way to broaden participation in computer architecture research and have been shown to improve student learning, engagement, and retention. These outcomes can be more profound and lasting if students experience research early. However, there are many barriers to early research in computer architecture some of which include the gap between pedagogy and research, the lower emphasis on hardware design compared to software in first-year courses, and the lack of online resources. We propose lowering these barriers through a methodical approach by involving undergraduates in early research and by creating freely available and innovative educational tools for designing hardware. We present the experience of a team of undergraduate students with research over one academic year using a Python hardware description language, PyRTL. PyRTL was developed to enable early entry into digital design. Its overarching goals are simplicity, usabil- ity, clarity, and extensibility, a stark contrast to traditional languages like Verilog and VHDL that have a steep learning curve. Instead of introducing traditional languages early in the undergraduate curriculum, PyRTL takes the opposite approach, which is to build on what students already know well: a popular programming language (Python), software design patterns, and software engineering principles. The students conducted their research in the context of the Early Research Scholars Program (ERSP), a program designed to expand access to research among women and underrepresented minority students in their second year through a well-designed support structure. 

As Machine Learning (ML) applications become pervasive and computer architects further integrate hardware support, the need to rapidly explore trade-offs between algorithms and hardware becomes pressing. While prior work on hardware accelerators has led to tremendous performance and energy improvements, it can be difficult to generalize these approaches without resorting to special-purpose tools or even languages. Through object-oriented design principles, we describe a general and reusable approach for generating parameterized neural network hardware. Specifically, we describe our experiences with high-level hardware design objects for building neural network hardware based on the open-source Python HDL, PyRTL. By thinking at a higher level of abstraction than simple “hardware modules,”, we open the door to a process by which hardware can be developed with software engineering principles. This creates new opportunities for a tight feedback loop between machine learning algorithm innovation and hardware design reality. Future works considering hardware development for ML applications can benefit from our work analyzing the costs and benefits of abstraction. 

Recent developments in Neural Relation Extraction (NRE) have made significant strides towards Automated Knowledge Base Construction. While much attention has been dedicated towards improvements in accuracy, there have been no attempts in the literature to evaluate social biases exhibited in NRE systems. In this paper, we create WikiGenderBias, a distantly supervised dataset composed of over 45,000 sentences including a 10% human annotated test set for the purpose of analyzing gender bias in relation extraction systems. We find that when extracting spouse-of and hypernym (i.e., occupation) relations, an NRE system performs differently when the gender of the target entity is different. However, such disparity does not appear when extracting relations such as birthDate or birthPlace. We also analyze how existing bias mitigation techniques, such as name anonymization, word embedding debiasing, and data augmentation affect the NRE system in terms of maintaining the test performance and reducing biases. Unfortunately, due to NRE models rely heavily on surface level cues, we find that existing bias mitigation approaches have a negative effect on NRE. Our analysis lays groundwork for future quantifying and mitigating bias in NRE. 

As Natural Language Processing (NLP) and Machine Learning (ML) tools rise in popularity, it becomes increasingly vital to recognize the role they play in shaping societal biases and stereotypes. Although NLP models have shown success in modeling various applications, they propagate and may even amplify gender bias found in text corpora. While the study of bias in artificial intelligence is not new, methods to mitigate gender bias in NLP are relatively nascent. In this paper, we review contemporary studies on recognizing and mitigating gender bias in NLP. We discuss gender bias based on four forms of representation bias and analyze methods recognizing gender bias. Furthermore, we discuss the advantages and drawbacks of existing gender debiasing methods. Finally, we discuss future studies for recognizing and mitigating gender bias in NLP.