A5100783908- Natural Language Processing Techniques
- Topic Modeling
- Multimodal Machine Learning Applications
- Text Readability and Simplification
- Innovative Teaching and Learning Methods
- Speech and dialogue systems
- Machine Learning and Data Classification
- Online and Blended Learning
- Emotional Intelligence and Performance
- Online Learning and Analytics
- Leadership, Courage, and Heroism Studies
- Education and Critical Thinking Development
- Personality Traits and Psychology
- Expert finding and Q&A systems
- Facilities and Workplace Management
- Speech Recognition and Synthesis
- Complex Network Analysis Techniques
- Student Assessment and Feedback
- Posttraumatic Stress Disorder Research
- Reproductive Health and Technologies
- EFL/ESL Teaching and Learning
- Machine Learning and Algorithms
- Structural Analysis of Composite Materials
- Neural Networks and Applications
- Building materials and conservation
University of Washington
2020-2023
New York University
2018-2022
Stanford University
2021
National Research Council Canada
2020
Nanyang Technological University
2016-2020
Technical University of Darmstadt
2020
Meta (Israel)
2020
Allen Institute
2020
Institut national de recherche en informatique et en automatique
2020
George Washington University
2020
In the last year, new models and methods for pretraining transfer learning have driven striking performance improvements across a range of language understanding tasks. The GLUE benchmark, introduced little over one year ago, offers single-number metric that summarizes progress on diverse set such tasks, but benchmark has recently surpassed level non-expert humans, suggesting limited headroom further research. this paper we present SuperGLUE, styled after with more difficult software...
Contextualized representation models such as ELMo (Peters et al., 2018a) and BERT (Devlin 2018) have recently achieved state-of-the-art results on a diverse array of downstream NLP tasks. Building recent token-level probing work, we introduce novel edge task design construct broad suite sub-sentence tasks derived from the traditional structured pipeline. We probe word-level contextual representations four investigate how they encode sentence structure across range syntactic, semantic, local,...
We show that BERT (Devlin et al., 2018) is a Markov random field language model. This formulation gives way to natural procedure sample sentences from BERT. generate and find it can produce high quality, fluent generations. Compared the generations of traditional left-to-right model, generates are more diverse but slightly worse quality.
We show that BERT (Devlin et al., 2018) is a Markov random field language model. This formulation gives way to natural procedure sample sentences from BERT. generate and find it can produce high-quality, fluent generations. Compared the generations of traditional left-to-right model, generates are more diverse but slightly worse quality.
Najoung Kim, Roma Patel, Adam Poliak, Patrick Xia, Alex Wang, Tom McCoy, Ian Tenney, Alexis Ross, Tal Linzen, Benjamin Van Durme, Samuel R. Bowman, Ellie Pavlick. Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM 2019). 2019.
Undirected neural sequence models such as BERT (Devlin et al., 2019) have received renewed interest due to their success on discriminative natural language understanding tasks question-answering and inference. The problem of generating sequences directly from these has relatively little attention, in part because undirected departs significantly conventional monotonic generation directed models. We investigate this by proposing a generalized model that unifies decoding proposed framework the...
Julian Michael, Ari Holtzman, Alicia Parrish, Aaron Mueller, Alex Wang, Angelica Chen, Divyam Madaan, Nikita Nangia, Richard Yuanzhe Pang, Jason Phang, Samuel R. Bowman. Proceedings of the 61st Annual Meeting Association for Computational Linguistics (Volume 1: Long Papers). 2023.
Sebastian Gehrmann, Abhik Bhattacharjee, Abinaya Mahendiran, Alex Wang, Alexandros Papangelis, Aman Madaan, Angelina Mcmillan-major, Anna Shvets, Ashish Upadhyay, Bernd Bohnet, Bingsheng Yao, Bryan Wilie, Chandra Bhagavatula, Chaobin You, Craig Thomson, Cristina Garbacea, Dakuo Daniel Deutsch, Deyi Xiong, Di Jin, Dimitra Gkatzia, Dragomir Radev, Elizabeth Clark, Esin Durmus, Faisal Ladhak, Filip Ginter, Genta Indra Winata, Hendrik Strobelt, Hiroaki Hayashi, Jekaterina Novikova, Jenna...
Asylum-seekers from Africa immigrate to Israel through the Sinai desert and are often exposed traumatic events.To identify scope types of medical services required by asylum-seekers relationship between delayed care development post-traumatic stress disorder (PTSD) overutilization services.Asylum-seekers that entered 2009 2012 who utilized Open Clinic Physicians for Human Rights were interviewed record their experiences in Sinai, document events they to, diagnoses, clinic visits. Linkages...
Understanding the ways in which users interact with different online communities is crucial to social network analysis and community maintenance.We present an unsupervised neural model learn linguistic descriptors for a user's behavior over time within community.We show that learned by our capture functional roles occupy communities, contrast those via standard topic-modeling algorithm, simply reflect topical content.Experiments on media forum Reddit how can provide interpretable insights...
Large volumes of text data have contributed significantly to the development large language models (LLMs) in recent years. This is typically acquired by scraping internet, leading pretraining datasets comprised noisy web text. To date, efforts prune these down a higher quality subset relied on hand-crafted heuristics encoded as rule-based filters. In this work, we take wider view and explore scalable estimates that can be used systematically measure data. We perform rigorous comparison at...
We present the results of NLP Community Metasurvey. Run from May to June 2022, survey elicited opinions on controversial issues, including industry influence in field, concerns about AGI, and ethics. Our put concrete numbers several controversies: For example, respondents are split almost exactly half questions importance artificial general intelligence, whether language models understand language, necessity linguistic structure inductive bias for solving problems. In addition, posed...
The Word Embedding Association Test shows that GloVe and word2vec word embeddings exhibit human-like implicit biases based on gender, race, other social constructs (Caliskan et al., 2017). Meanwhile, research learning reusable text representations has begun to explore sentence-level texts, with some sentence encoders seeing enthusiastic adoption. Accordingly, we extend the measure bias in encoders. We then test several encoders, including state-of-the-art methods such as ELMo BERT, for...
We introduce a set of nine challenge tasks that test for the understanding function words. These are created by structurally mutating sentences from existing datasets to target comprehension specific types words (e.g., prepositions, wh-words). Using these probing tasks, we explore effects various pretraining objectives sentence encoders language modeling, CCG supertagging and natural inference (NLI)) on learned representations. Our results show modeling performs best average across our...
Evaluation in machine learning is usually informed by past choices, for example which datasets or metrics to use. This standardization enables the comparison on equal footing using leaderboards, but evaluation choices become sub-optimal as better alternatives arise. problem especially pertinent natural language generation requires ever-improving suites of datasets, metrics, and human make definitive claims. To following best model practices easier, we introduce GEMv2. The new version...
Abstract: Committing to a major is fateful step in an undergraduate education, yet the relationship between courses taken early academic career and ultimate selection remains little studied at scale. Using transcript data capturing careers of 26,892 undergraduates enrolled private university 2000 2020, we describe enrollment histories using natural-language methods vector embeddings forecast terminal on basis course sequences beginning college entry. We find (I) student's very first predicts...
You have accessJournal of UrologyCME1 Apr 2023MP02-11 ANDROGEN AND ESTROGEN RECEPTOR CHANGES IN SCROTAL SKIN OF TRANSGENDER PATIENTS UNDERGOING GENDER AFFIRMING SURGERY Michele Fascelli, Matthew Bury, Alex Wang, Matthias Hofer, and Daniel Dugi FascelliMichele Fascelli More articles by this author , BuryMatthew Bury WangAlex Wang HoferMatthias Hofer DugiDaniel View All Author Informationhttps://doi.org/10.1097/JU.0000000000003213.11AboutPDF ToolsAdd to favoritesDownload CitationsTrack...