Title
Topic
-
‘On Centralized Critics in Multi-Agent Reinforcement Learning’
“Centralized Training for Decentralized Execution, where agents are trained offline in a centralized fashion and execute online in a decentralized manner, has become a popular approach in Multi-Agent Reinforcement Learning (MARL). In particular, it has become popular to develop actor-critic methods that train decentralized actors with a centralized critic … [however,] using a centralized critic in this context has yet to be sufficiently analyzed theoretically or empirically. In this paper, we therefore formally analyze centralized and decentralized critic approaches.” Find the paper and full list of authors at the Journal of Artificial Intelligence Research.
-
‘Discovering Variable Binding Circuitry With Desiderata’
“Recent work has shown that computation in language models may be human-understandable, with successful efforts to localize and intervene on both single-unit features and input-output circuits. Here, we introduce an approach which extends causal mediation experiments to automatically identify model components responsible for performing a specific subtask by solely specifying a set of desiderata, or causal attributes of the model components executing that subtask.” Find the paper and full list of authors at ArXiv.
-
‘Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task’
“Language models show a surprising range of capabilities, but the source of their apparent competence is unclear. Do these networks just memorize a collection of surface statistics, or do they rely on internal representations of the process that generates the sequences they see? We investigate this question in a synthetic setting by applying a variant of the GPT model to the task of predicting legal moves in a simple board game, Othello.” Find the paper and full list of authors at Open Review.
-
‘Mass-Editing Memory in a Transformer’
“Recent work has shown exciting promise in updating large language models with new memories, so as to replace obsolete information or add specialized knowledge. However, this line of work is predominantly limited to updating single associations. We develop MEMIT, a method for directly updating a language model with many memories, demonstrating experimentally that it can scale up to thousands of associations for GPT-J (6B) and GPT-NeoX (20B), exceeding prior work by orders of magnitude.” Find the paper and full list of authors at Open Review.
-
Ramdin delivers lecture on mentorship and excellence at ABNF
Valeria Ramdin, associate clinical professor and director of global health nursing, presented a talk on “Mentoring BIPOC Nursing Faculty Toward Leadership Excellence: A Concept Analysis With Historical Research” at the 35th Annual Meeting & Scientific Conference for the Association of Black Nursing Faculty.
-
‘Environmental and Geographical Factors Structure Cauliflower Coral’s Algal Symbioses Across the Indo-Pacific’
“The symbioses between corals and endosymbiotic dinoflagellates have been described as a flexible relationship whose dynamics could serve as a source of resilience for coral reef ecosystems. However, the factors that drive the establishment and maintenance of this co-evolutionary relationship remain unclear. We examined the environmental and geographical factors structuring dinoflagellate communities in a wide-ranging Indo-Pacific coral to begin to address this gap. … We provide further support for the hypothesis that coral’s Symbiodiniaceae communities could facilitate host resilience to thermal stress.” Find the paper and full list of authors at the Journal of Biogeography.
-
Northeastern University launches fully automated and virtualized O-RAN private 5G network with AI automation
“The Institute for the Wireless Internet of Things (WIoT) at Northeastern University and its Open6G R&D Center announce the availability of the first production-ready private 5G network fully automated through Artificial Intelligence (AI). The system is built on open-source components enabling a fully virtualized, programmable O-RAN compliant network in a campus environment.”
-
Chowdhury selected as finalist for Blavatnik National Award in Physical Sciences & Engineering
“Electrical and computer engineering professor Kaushik Chowdhury was selected as a finalist for the 2023 Blavatnik National Awards for Young Scientists in Physical Sciences & Engineering for addressing the global need of telecommunications spectrum scarcity, as well as improve connectivity by designing next generation wireless systems and machine learning-based network operations. The prestigious Blavatnik National Awards for Young Scientists is the largest unrestricted prize for early career scientists and honors outstanding young scientists and engineers under the age of 42.”
-
‘Semantics-Aware Dataset Discovery From Data Lakes With Contextualized Column-Based Representation Learning’
“Dataset discovery from data lakes is essential in many real application scenarios. In this paper, we propose Starmie, an end-to-end framework for dataset discovery from data lakes (with table union search as the main use case). Our proposed framework features a contrastive learning method to train column encoders from pre-trained language models in a fully unsupervised manner. The column encoder of Starmie captures the rich contextual semantic information within tables by leveraging a contrastive multi-column pre-training strategy.” Find the paper and the full list of authors in the Proceedings of the VLDB Endowment.
-
‘Explaining Dataset Changes for Semantic Data Versioning With Explain-Da-V’
“In multi-user environments in which data science and analysis is collaborative, multiple versions of the same datasets are generated. While managing and storing data versions has received some attention in the research literature, the semantic nature of such changes has remained under-explored. In this work, we introduce Explain-Da-V, a framework aiming to explain changes between two given dataset versions. Explain-Da-V generates explanations that use data transformations to explain changes. We further introduce a set of measures that evaluate the validity, generalizability, and explainability of these explanations.” Find the paper and full list of authors in VLDB Endowment proceedings.
-
‘Table Discovery in Data Lakes: State-of-the-Art and Future Directions’
“Data discovery refers to a set of tasks that enable users and downstream applications to explore and gain insights from massive collections of data sources such as data lakes. In this tutorial, we will provide a comprehensive overview of the most recent table discovery techniques developed by the data management community. We will cover table understanding tasks such as domain discovery, table annotation, and table representation learning which help data lake systems capture semantics of tables.” Find the paper and the full list of authors in the Companion of the 2023 International Conference on Management of Data.
-
‘SANTOS: Relationship-Based Semantic Table Union Search’
“Existing techniques for unionable table search define unionability using metadata (tables must have the same or similar schemas) or column-based metrics (for example, the values in a table should be drawn from the same domain). In this work, we introduce the use of semantic relationships between pairs of columns in a table to improve the accuracy of the union search. Consequently, we introduce a new notion of unionability that considers relationships between columns, together with the semantics of columns, in a principled way.” Find the paper and full list of authors in the Proceedings of ACM on Management of Data.
-
‘Direct Superpoints Matching for Fast and Robust Point Cloud Registration’
“Although deep neural networks endow the downsampled superpoints with discriminative feature representations, directly matching them is usually not used alone in state-of-the-art methods. … Existing approaches use the coarse-to-fine strategy to propagate the superpoints correspondences to the point level, which are not discriminative enough and further necessitates the postprocessing refinement. In this paper, we present a simple yet effective approach to extract correspondences by directly matching superpoints using a global softmax layer in an end-to-end manner, which are used to determine the rigid transformation between the source and target point cloud.” Find the paper and full list of authors at…
-
‘Do Machine Learning Models Produce TypeScript Types That Type Check? (Artifact)’
“Type migration is the process of adding types to untyped code to gain assurance at compile time. TypeScript and other gradual type systems facilitate type migration by allowing programmers to start with imprecise types and gradually strengthen them. … In this paper we argue that accuracy can be misleading, and we should address a different question: can an automatic type migration tool produce code that passes the TypeScript type checker? We present TypeWeaver, a TypeScript type migration tool that can be used with an arbitrary type prediction model.” Find the paper and full list of authors at Dagstuhl Research Online…
-
‘Online Learning in Multi-Unit Auctions’
“We consider repeated multi-unit auctions with uniform pricing, which are widely used in practice for allocating goods such as carbon licenses. In each round, K identical units of a good are sold to a group of buyers that have valuations with diminishing marginal returns. The buyers submit bids for the units, and then a price p is set per unit so that all the units are sold. We consider two variants of the auction, where the price is set to the K-th highest bid and (K+1)-st highest bid, respectively.” Find the paper and full list of authors at ArXiv.
-
‘State of the Art of Visual Analytics for eXplainable Deep Learning’
“The use and creation of machine-learning-based solutions to solve problems or reduce their computational costs are becoming increasingly widespread in many domains. … This survey aims to (i) systematically report the contributions of Visual Analytics for eXplainable Deep Learning; (ii) spot gaps and challenges; (iii) serve as an anthology of visual analytical solutions ready to be exploited and put into operation by the Deep Learning community (architects, trainers and end users) and (iv) prove the degree of maturity, ease of integration and results for specific domains.” Find the paper and full list of authors at Computer Graphics Forum.
-
Libby awarded $1.96M Early-Stage Investigator Grant from NIH
“Elizabeth Libby, assistant professor of bioengineering, recently received a five-year, $1.96 million Early Stage Investigator R35 MIRA (Maximizing Investigator’s Research Award) grant from the National Institutes of Health for ‘Physiological and Developmental Role of Bacterial Ser/Thr Kinases.’ Libby’s research is focused on how bacteria develop resistance at the cellular level—knowledge that will be crucial to the development of more effective antibiotics.”
-
Rouhanifard receives $3.4M NIH grant for modified mRNA research
“Bioengineering assistant professor Sara Rouhanifard was awarded a $3.4 million NIH R01 grant for ‘Synthetic mRNA Control Set for Nanopore-Based Pseudouridine Modification Profiling in Human Transcriptomes.’ The research has the potential to vastly increase insight into the epitranscriptome—changes in chemical modifications of RNA that can affect gene expression within cells—which could help identify new therapeutic targets and lead to new classes of drugs.”
-
‘Adapting Transformer Language Models for Predictive Typing in Brain-Computer Interfaces’
“Brain-computer interfaces (BCI) are an important mode of alternative and augmentative communication for many people. Unlike keyboards, many BCI systems do not display even the 26 letters of English at one time, let alone all the symbols in more complex systems. Using language models to make character-level predictions, therefore, can greatly speed up BCI typing (Ghosh and Kristensson, 2017). While most existing BCI systems employ character n-gram models or no LM at all, this paper adapts several wordpiece-level Transformer LMs to make character predictions and evaluates them on typing tasks.” Find the paper and the full list of authors at ArXiv.
-
‘Composition and Deformance: Measuring Imageability With a Text-to-Image Model’
“Although psycholinguists and psychologists have long studied the tendency of linguistic strings to evoke mental images in hearers or readers, most computational studies have applied this concept of imageability only to isolated words. Using recent developments in text-to-image generation models, such as DALLE mini, we propose computational methods that use generated images to measure the imageability of both single English words and connected text. We sample text prompts for image generation … [and] subject these prompts to different deformances to examine the model’s ability to detect changes in imageability caused by compositional change.” Find the paper and full authors list…
-
‘Adaptively Secure MPC With Sublinear Communication Complexity’
“A central challenge in the study of MPC is to balance between security guarantees, hardness assumptions, and resources required for the protocol. In this work, we study the cost of tolerating adaptive corruptions in MPC protocols under various corruption thresholds. … Our results highlight that the asymptotic cost of adaptive security can be reduced to be comparable to, and in many settings almost match, that of static security, with only a little sacrifice to the concrete round complexity and asymptotic communication complexity.” Find the paper and the full list of authors at the Journal of Cryptology.
-
‘Threshold ECDSA in Three Rounds’
” “We present a three-round protocol for threshold ECDSA signing with malicious security against a dishonest majority, which information-theoretically UC-realizes a standard threshold signing functionality, assuming ideal commitment and two-party multiplication primitives. Our work improves upon and fully subsumes the DKLs t-of-n and 2-of-n protocols. This document focuses on providing a succinct but complete description of the protocol and its security proof, and contains little expository text.” Find the paper and the full list of authors at the Cryptology ePrint Archive.
-
‘Threshold BBS+ Signatures for Distributed Anonymous Credential Issuance’
“We propose a secure multiparty signing protocol for the BBS+ signature scheme; in other words, an anonymous credential scheme with threshold issuance. We prove that due to the structure of the BBS+ signature, simply verifying the signature produced by an otherwise semi-honest protocol is sufficient to achieve composable security against a malicious adversary. Consequently, our protocol is extremely simple and efficient: it involves a single request from the client … to the signing parties, two exchanges of messages among the signing parties, and finally a response to the client.” Find the paper and full list of authors at Cryptology ePrint…
-
‘JaX: Detecting and Cancelling High-Power Jammers Using Convolutional Neural Network’
“We present JaX, a novel approach for detecting and cancelling high-power jammers in the scenarios when the traditional spread spectrum techniques and other jammer avoidance approaches are not sufficient. JaX does not require explicit probes, sounding, training sequences, channel estimation, or the cooperation of the transmitter. We identify and address multiple challenges, resulting in a convolutional neural network for a multi-antenna system to infer the existence of interference, the number of interfering emissions and their respective phases.” Find the paper and full list of authors in the 16th ACM Conference on Security and Privacy in Wireless and Mobile Networks proceedings.
-
‘A Formal Analysis of Karn’s Algorithm’
“The stability of the Internet relies on timeouts. The timeout value, known as the Retransmission TimeOut (RTO), is constantly updated, based on sampling the Round Trip Time (RTT) of each packet as measured by its sender – that is, the time between when the sender transmits a packet and receives a corresponding acknowledgement. Many of the Internet protocols compute those samples via the same sampling mechanism, known as Karn’s Algorithm. We present a formal description of the algorithm, and study its properties.” Find the paper and the full list of authors in Networked Systems.
-
‘Evaluating the Impact of Community Oversight for Managing Mobile Privacy and Security’
“Mobile privacy and security can be a collaborative process where individuals seek advice and help from their trusted communities. To support such collective privacy and security management, we developed a mobile app for Community Oversight of Privacy and Security (“CO-oPS”) that allows community members to review one another’s apps installed and permissions granted to provide feedback. … Measures of transparency, trust, and awareness of one another’s mobile privacy and security behaviors, along with individual and community participation in mobile privacy and security co-management, increased from pre- to post-study.” Find the paper and the full list of authors at ArXiv.
-
‘Disaster World: Decision-Theoretic Agents for Simulating Population Responses to Hurricanes’
“Artificial intelligence (AI) research provides a rich source of modeling languages capable of generating socially plausible simulations of human behavior, while also providing a transparent ground truth that can support validation of social-science methods applied to that simulation. In this work, we leverage two established AI representations: decision-theoretic planning and recursive modeling. … We used PsychSim, a multiagent social-simulation framework combining these two AI frameworks, to build a general parameterized model of human behavior during disaster response, grounding the model in social-psychological theories to ensure social plausibility.” Find the paper and the full list of authors at Computational and Mathematical Organization…
-
‘Effectiveness of Teamwork-Level Interventions Through Decision-Theoretic Reasoning in a Minecraft Search-and-Rescue Task’
“Autonomous agents offer the promise of improved human teamwork through automated assessment and assistance during task performance. Studies of human teamwork have identified various processes that underlie joint task performance, while abstracting away the specifics of the task.We present here an agent that focuses exclusively on teamwork-level variables in deciding what interventions to use in assisting a human team. Our agent … relies on input from analytic components (ACs) (developed by other research teams) that process environmental information and output only teamwork-relevant measures.” Find the paper and authors list in the 2023 International Conference on Autonomous Agents and Multiagent Systems…
-
‘Agent-Based Modeling of Human Decision-Makers Under Uncertain Information During Supply Chain Shortages’
“In recent years, product shortages caused by supply chain disruptions have generated problems for consumers worldwide. … Understanding how humans learn to interpret information from others and how it influences their decision-making is key to alleviating supply chain shortages. In this work, we investigated how downstream supply chain echelons, health centers in pharmaceutical supply chains, interpret and use manufacturers’ estimated resupply date (ERD) information during drug shortages.” Find the paper and the full list of authors in the 2023 International Conference on Autonomous Agents and Multiagent Systems proceedings.