Title
Topic
-
‘Linearity of Relation Decoding in Transformer Language Models’
“Much of the knowledge encoded in transformer language models (LMs) may be expressed in terms of relations: relations between words and their synonyms, entities and their attributes, etc. We show that, for a subset of relations, this computation is well-approximated by a single linear transformation on the subject representation. Linear relation representations may be obtained by constructing a first-order approximation to the LM from a single prompt, and they exist for a variety of factual, commonsense, and linguistic relations.” Find the paper and full list of authors at ArXiv.
-
‘TikTok as Algorithmically Mediated Biographical Illumination: Autism, Self-Discovery and Platformed Diagnosis on #Autisktok’
“Scholarship in the sociology of medicine has tended to characterize diagnosis as disruptive to one’s self-concept. This categorization, though, requires reconsideration in light of public conversations about mental health and community building around neurocognitive conditions, particularly among youth online. … We explored the shifting nature of [‘biographical illumination’] through the case of TikTok. Combining quantitative and qualitative methods, we argue that TikTok serves as a space to discuss diagnosis and refine one’s sense of self as a result of diagnosis.” Find the paper and full list of authors at New Media & Society.
-
‘Why Look at Dead Animals?’ asks Coughlin in provocative taxidermy study
“Lion Attacking a Dromedary was a sensational object for its first viewers at the Paris Universal Exposition in 1867. … As we now know, the Verreaux brothers embedded human remains in the figure of the rider that had formerly been assumed to be just a clothed mannequin. … This essay suggests that theoretical tools derived from Material Ecocriticism and Monster Theory that may help us to think about, or alongside, the affective power of this disturbing taxidermy assemblage, ever aware that this piece draws its power from the theatrical, colonial violence of extraction and extinction.”
-
‘Deep Bayesian Active Learning for Accelerating Stochastic Simulation’
“Stochastic simulations such as large-scale, spatiotemporal, age-structured epidemic models are computationally expensive at fine-grained resolution. While deep surrogate models can speed up the simulations, doing so for stochastic simulations and with active learning approaches is an underexplored area. We propose Interactive Neural Process (INP), a deep Bayesian active learning framework for learning deep surrogate models to accelerate stochastic simulations. INP consists of two components, a spatiotemporal surrogate model built upon Neural Process (NP) family and an acquisition function for active learning.” Find the paper and full list of authors in the SIGKDD Conference on Knowledge Discovery and Data Mining proceedings.
-
‘Disentangling Node Attributes From Graph Topology for Improved Generalizability in Link Prediction’
“Link prediction is a crucial task in graph machine learning with diverse applications. We explore the interplay between node attributes and graph topology and demonstrate that incorporating pre-trained node attributes improves the generalization power of link prediction models. Our proposed method, UPNA (Unsupervised Pre-training of Node Attributes), solves the inductive link prediction problem by learning a function that takes a pair of node attributes and predicts the probability of an edge, as opposed to Graph Neural Networks, … which can be prone to topological shortcuts in graphs with power-law degree distribution.” Find the paper and full list of authors at…
-
‘Can Euclidean Symmetry Be Leveraged in Reinforcement Learning and Planning?’
“In robotic tasks, changes in reference frames typically do not influence the underlying physical properties of the system, which has been known as invariance of physical laws.These changes, which preserve distance, encompass isometric transformations such as translations, rotations, and reflections, collectively known as the Euclidean group. In this work, we delve into the design of improved learning algorithms for reinforcement learning and planning tasks that possess Euclidean group symmetry.” Find the paper and full list of authors at ArXiv.
-
‘Leveraging Structure for Improved Classification of Grouped Biased Data’
“We consider semi-supervised binary classification for applications in which data points are naturally grouped … and the labeled data is biased. … The groups overlap in the feature space and consequently the input-output patterns are related across the groups. To model the inherent structure in such data, we assume the partition-projected class-conditional invariance across groups. … We demonstrate that under this assumption, the group carries additional information about the class, over the group-agnostic features, with provably improved area under the ROC curve.” Find the paper and full list of authors in the AAAI Conference on Artificial Intelligence proceedings.
-
‘Accelerating Neural MCTS Algorithms Using Neural Sub-Net Structures’
“Neural MCTS algorithms are a combination of Deep Neural Networks and Monte Carlo Tree Search (MCTS) and have successfully trained Reinforcement Learning agents in a tabula-rasa way. … However, these algorithms … take a long time to converge, which requires high computational power and electrical energy. It also becomes difficult for researchers without cutting-edge hardware to pursue Neural MCTS research. We propose Step-MCTS, a novel algorithm that uses subnet structures, each of which simulates a tree that provides a lookahead for exploration.” Find the paper and full list of authors in the International Conference on Autonomous Agents and Multiagent Systems…
-
‘Sustainable HCI Under Water: Opportunities for Research with Oceans, Coastal Communities and Marine Systems’
“Although the world’s oceans play a critical role in human well-being, they have not been a primary focus of the sustainable HCI (SHCI) community to date. In this paper, we present a scoping review to show how concerns with the oceans are threaded throughout the broader SHCI literature and to find new research opportunities. We identify several themes that could benefit from focused SHCI research, including marine food sources, culture and coastal communities, ocean conservation, and marine climate change impacts and adaptation strategies.” Find the paper and full list of authors at the Conference on Human Factors in Computing Systems.
-
‘Rapid Convergence: The Outcomes of Making PPE During a Healthcare Crisis’
“The U.S. National Institute of Health (NIH) 3D Print Exchange is a public, open-source repository for 3D printable medical device designs with contributions from clinicians, expert-amateur makers, and people from industry and academia. In response to the COVID-19 pandemic, the NIH formed a collection to foster submissions of low-cost, locally manufacturable personal protective equipment (PPE). We evaluated the 623 submissions in this collection … [and] found an immediate design convergence to manufacturing-focused remixes of a few initial designs affiliated with NIH partners and major for-profit groups.” Find the paper and full list of authors at ACM Transactions on Computer-Human Interaction.
-
‘OPTIMISM: Enabling Collaborative Implementation of Domain Specific Metaheuristic Optimization’
“For non-technical domain experts and designers it can be a substantial challenge to create designs that meet domain specific goals. This presents an opportunity to create specialized tools that produce optimized designs in the domain. However, implementing domain-specific optimization methods requires a rare combination of programming and domain expertise. … We present OPTIMISM, a toolkit which enables programmers and domain experts to collaboratively implement an optimization component of design tools.” Find the paper and full list of authors in the Conference on Human Factors in Computing Systems proceedings.
-
Understanding human decision-making during supply chains shortages
“Research conducted by mechanical and industrial engineering associate professor Jacqueline Griffin, professor Ozlem Ergun, and professor Stacy Marsella [in the Khoury College of Computer science, titled] ‘Agent-Based Modeling of Human Decision-Makers Under Uncertain Information During Supply Chain Shortages’ was published in the proceedings from the 2023 International Conference on Autonomous Agents and Multiagent Systems.”
-
‘Predictors and Consequences of Pro-Environmental Behavior at Work’
“Increasingly, people are looking for meaning through their jobs, for employers that have a positive impact on the world, and for workplaces that promote mission-driven behavior. One such mission that is a growing priority is addressing climate change, especially for younger cohorts entering the workforce. Addressing the climate crisis will necessitate substantial changes at all levels of society, including organizational change. This paper examines individual, social, and contextual variables that are associated with pro-environmental behavior (PEB).” Find the paper and full list of authors at ResearchGate.
-
‘Microbial Chemolithoautotrophs are Abundant in Salt Marsh Sediment Following Long-Term Experimental Nitrate Enrichment’
“Long-term anthropogenic nitrate (NO3−) enrichment is a serious threat to many coastal systems. Nitrate reduction coupled with the oxidation of reduced forms of sulfur is conducted by chemolithoautotrophic microbial populations in a process that decreases nitrogen (N) pollution. However, little is known about the diversity and distribution of microbes capable of carbon fixation within salt marsh sediment and how they respond to long-term NO3− loading. We used genome-resolved metagenomics to characterize the distribution, phylogenetic relationships, and adaptations important to microbial communities within NO3− enriched sediment.” Find the paper and full list of authors at FEMS Microbiology Letters.
-
‘Workforce Ecosystems and AI’
“Companies increasingly rely on an extended workforce (e.g., contractors, gig workers … and technologies such as algorithmic management and artificial intelligence) to achieve strategic goals and objectives. When we ask leaders to describe how they define their workforce today, they mention a diverse array of participants, beyond just full- and part-time employees, all contributing in various ways. … Our ongoing research on workforce ecosystems demonstrates that managing work across organizational boundaries with groups of interdependent actors in a variety of employment relationships creates new opportunities and risks for both workers and businesses.” Find the paper and full list of authors…
-
‘A Tool for Mutation Analysis in Racket’
“Racket is a functional programming language that is used to teach CS1 at many high schools and colleges. … In order to evaluate [mutation analysis’s] efficacy in our college’s introductory programming courses, we created a prototype mutation analysis tool for Racket. We describe the design and features of the tool and perform a feasibility study using two assignments from an intro CS course where student test suite thoroughness was evaluated by hand by human graders.” Find the paper and full list of authors in the 2023 IEEE International Conference on Software Testing Verification and Validation Workshop.
-
‘SliceLens: Guided Exploration of Machine Learning Datasets’
“SliceLens is a tool for exploring labeled, tabular, machine learning datasets. To explore a dataset, the user selects combinations of features in the dataset that they are interested in. The tool splits those features into bins and then visualizes the label distributions for the subsets of data created by the intersections of the bins. SliceLens guides the user in determining which feature combinations to explore. Guidance is based on a user-selected rating metric, which assigns a score to the subsets created by a given combination of features.”