Emily Cheng @ ACL 2025
@sparse_emcheng
Apple MLR intern and PhD @colt_upf in computational linguistics What is the happiest state? Maryland 💅🏼 Before: MIT CSAIL, ENS
Uncertainty quantification (UQ) is key for safe, reliable LLMs... but are we evaluating it correctly? 🚨 Our ACL2025 paper finds a hidden flaw: if both UQ methods and correctness metrics are biased by the same factor (e.g., response length), evaluations get systematically skewed
A personal update. bhaskar-mitra.github.io/posts/2025/07/…
PostDoc position (3.5y) available in interdisciplinary Collaborative Research Center "Common Ground". 🎯 Topic: Pragmatic reasoning about Common Ground 🎓 linguistics | philosophy of language | cognitive science 📅 Deadline: July 14 tinyurl.com/3vd9wa6p Please share!
🚨New Preprint!! Thrilled to share with you our latest work: “Mixture of Cognitive Reasoners”, a modular transformer architecture inspired by the brain’s functional networks: language, logic, social reasoning, and world knowledge. 1/ 🧵👇
This paper is a bit heavy; however, the insights are good. The authors characterize the requirements for two representation to be similar. Also, there may be a connection with openreview.net/forum?id=yyYMA… (I should look into this). 🔗arxiv.org/abs/2506.03784
I'm so glad this paper finally lost the "release date procrastination" race against GTA6! 🥹
Neural networks implicitly define a latent vector field on the data manifold, via autoencoding iterations🌀 This representation retains properties of the model, revealing memorization and generalization regimes, and characterizing distribution shifts 📜: arxiv.org/abs/2505.22785
Come find us in Hall 3! Can’t wait to meet you and talk about modeling brain data across heterogeneous datasets!
Want to scale models on brain datasets recorded with variable sensor layouts? Population Transformer at #ICLR2025 may be your answer! 🗺️ Fri, Apr 25 | 10am - 12:30pm (poster @ Hall 3 + Hall 2B #58) 🗣️ Fri, Apr 25 | 4:06 pm - 4:18 pm (oral @ Garnet 216-218) More ⬇️
Want to scale models on brain datasets recorded with variable sensor layouts? Population Transformer at #ICLR2025 may be your answer! 🗺️ Fri, Apr 25 | 10am - 12:30pm (poster @ Hall 3 + Hall 2B #58) 🗣️ Fri, Apr 25 | 4:06 pm - 4:18 pm (oral @ Garnet 216-218) More ⬇️
Brains, Minds and Machines Summer Course 2025. Application deadline: Mar 24, 2025 mbl.edu/education/adva… See more information here: cbmm.mit.edu/summer-school/…
🎉Excited to share: My first ML conference paper, Population Transformer 🧠, is an Oral at #ICLR2025! This work has truly evolved since its first appearance as a workshop paper last year. So thankful to have worked with the best advisors + collaborators! 🤗 More soon!
How can we train models on more brains and sensor layouts? We present Population Transformer (PopT) which learns population-level interactions on intracranial electrodes, with 🔥decoding and interpretability benefits. See our poster at #ICML2024 @AI_for_Science 12pm
Thank you so much to all the people that came to discuss our research! This paper was such a fun and exciting collaboration 🔥
If you're at #NeurIPS2024 today, check out our paper "Bridging semantics and pragmatics in information-theoretic emergent communication" neurips.cc/virtual/2024/p… Poster session: Fri Dec 13, 2-5pm, #3711 w/ @elegualdoni @roger_p_levy @MycalTucker
If you're at #NeurIPS2024 today, check out our paper "Bridging semantics and pragmatics in information-theoretic emergent communication" neurips.cc/virtual/2024/p… Poster session: Fri Dec 13, 2-5pm, #3711 w/ @elegualdoni @roger_p_levy @MycalTucker
I just landed in Vancouver to present @NeurIPSConf the findings of our new work! Few-shot learning and fine-tuning change the hidden layers of LLMs in a dramatically different way, even when they perform equally well on multiple-choice question-answering tasks. 🧵1/6
Why do NNs often learn similar representations? Existing identifiability results offer theoretical insights, but applying them in practice poses challenges. We’ll present our new work exploring these challenges next week at @unireps #NeurIPS2024 🇨🇦🎉 openreview.net/pdf?id=SQKUZSi… 1/
🔊New EMNLP paper from @elegualdoni & Gemma Boleda! Why do objects have many names? Human lexicons contain different words that speakers can use to refer to the same object, e.g., purple or magenta for the same color. We investigate using tools from efficient coding... 1/3
I'm recruiting PhD students this cycle! My lab works at the intersection of information theory, cognition, language, and AI. Wanna hear more? I asked notebookLM to generate a podcast just for you (but pls take it with a big grain of salt...) youtu.be/9L04S4jvspI
I’ll be presenting our work on predicting human reading comprehension from eye movements 👀🎯at #EMNLP2024! Tomorrow, Nov 12 (Tue) 16:00-17:30, Session 04 (Linguistic Theories, Cognitive Modeling and Psycholinguistics 2) Hope to see you there!
Is it possible to predict human reading comprehension solely form eye movements during reading? Check out our lab's EMNLP 2024 paper by @scaperex, @YoavMeiri and Cfir Hadar, that studies this long-standing question systematically. TLDR: it's very hard! arxiv.org/abs/2410.04484
New paper💡! Certain networks can't perform certain tasks due to lacking the right prior 😢. Can we make these untrainable networks trainable 🤔? We can, by introducing the prior through representational alignment with a trainable network! This approach is called guidance. (1/8)