About

Sang-Hwa Oh is an interdisciplinary researcher, teacher, and consultant specializing in societal- and individual-level wellbeing through innovative communications and media strategies. Her research investigates emerging media effects, the dissemination of health and risk misinformation, and how digital technologies can be leveraged to promote health prevention and the public good. Her work addresses the role of media and emerging communication technologies in shaping public understanding of urgent health and social issues, influencing behaviors at both individual and policy levels, strategies to mitigate misinformation, and how brands can enhance communication efforts to promote public good through transparency, trust, and emotional engagement. Dr. Oh has a background in mass communications, public health, and social welfare, with academic degrees from Ewha Womans University, Sogang University, and the University of South Carolina. She has served on editorial boards of prominent journals and her research has been published in esteemed outlets such as Health Communication, Risk Analysis, and the International Journal of Communication. Her interdisciplinary approach combines insights from communication, media effects, public relations, public health, psychology, sociology, and computational analysis to provide a comprehensive view of digital media effects and misinformation.

Research topics

• Computer science
• Artificial intelligence
• Mathematics
• Algorithm
• Machine learning

Selected publications

• PLeaS — Merging Models with Permutations and Least Squares

  2025-06-10 · 2 citations

  articleSenior author

  The democratization of machine learning systems has made the process of fine-tuning accessible to practitioners, leading to a wide range of open-source models fine-tuned on specialized tasks and datasets. Recent work has proposed to merge such models to combine their functionalities. However, prior approaches are usually restricted to models that are fine-tuned from the same base model. Furthermore, the final merged model is typically required to be of the same size as the original models. In this work, we propose a new two-step algorithm to merge models—termed PLeaS—which relaxes these constraints. First, leveraging the Permutation symmetries inherent in the two models, PLeaS partially matches nodes in each layer by maximizing alignment. Next, PLeaS computes the weights of the merged model as a layer-wise Least Squares solution to minimize the approximation error between the features of the merged model and the permuted features of the original models. PLeaS allows a practitioner to merge two models sharing the same architecture into a single performant model of a desired size, even when the two original models are fine-tuned from different base models. We also demonstrate how our method can be extended to address a challenging scenario where no data is available from the fine-tuning domains. We demonstrate our method to merge ResNet and ViT models trained with shared and different label spaces, and show improvement over the state-of-the-art merging methods of up to 15 percentage points for the same target compute while merging models trained on Domain-Net and fine-grained classification tasks<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>.

  PublisherDOI

• When Incentives Backfire, Data Stops Being Human

  ArXiv.org · 2025-02-11

  preprintOpen accessSenior author

  Progress in AI has relied on human-generated data, from annotator marketplaces to the wider Internet. However, the widespread use of large language models now threatens the quality and integrity of human-generated data on these very platforms. We argue that this issue goes beyond the immediate challenge of filtering AI-generated content -- it reveals deeper flaws in how data collection systems are designed. Existing systems often prioritize speed, scale, and efficiency at the cost of intrinsic human motivation, leading to declining engagement and data quality. We propose that rethinking data collection systems to align with contributors' intrinsic motivations -- rather than relying solely on external incentives -- can help sustain high-quality data sourcing at scale while maintaining contributor trust and long-term participation.

  PublisherOA PDFDOI

• Spurious Rewards: Rethinking Training Signals in RLVR

  arXiv (Cornell University) · 2025-06-12

  preprintOpen access

  We show that reinforcement learning with verifiable rewards (RLVR) can elicit strong mathematical reasoning in certain language models even with spurious rewards that have little, no, or even negative correlation with the correct answer. For example, RLVR training with GRPO improves MATH-500 performance for Qwen2.5-Math-7B by 21.4 percentage points using randomly assigned rewards, nearly matching the 29.1-point gain from ground-truth rewards. To explain this counterintuitive observation, we show that GRPO exhibits a clipping bias from the clip term, which can amplify high-prior behaviors learned during pretraining even without informative rewards. As a case study, we identify one such behavior in Qwen2.5-Math models, which we call code reasoning -- reasoning in code without actual code execution; code-reasoning frequency increases from 65 percent to over 90 percent with spurious rewards. However, the presence of such amplifiable behaviors is highly model-dependent. In practice, spurious rewards that are effective for Qwen models often fail to produce gains for other model families, such as Llama3 or OLMo2. Our results highlight the importance of validating RL methods across diverse models rather than relying on a single de facto choice: large gains can arise on Qwen models even from random rewards that do not reflect genuine capability improvements.

  PublisherOA PDFDOI

• CHANCERY: Evaluating Corporate Governance Reasoning Capabilities in Language Models

  ArXiv.org · 2025-06-05

  preprintOpen access

  Law has long been a domain that has been popular in natural language processing (NLP) applications. Reasoning (ratiocination and the ability to make connections to precedent) is a core part of the practice of the law in the real world. Nevertheless, while multiple legal datasets exist, none have thus far focused specifically on reasoning tasks. We focus on a specific aspect of the legal landscape by introducing a corporate governance reasoning benchmark (CHANCERY) to test a model's ability to reason about whether executive/board/shareholder's proposed actions are consistent with corporate governance charters. This benchmark introduces a first-of-its-kind corporate governance reasoning test for language models - modeled after real world corporate governance law. The benchmark consists of a corporate charter (a set of governing covenants) and a proposal for executive action. The model's task is one of binary classification: reason about whether the action is consistent with the rules contained within the charter. We create the benchmark following established principles of corporate governance - 24 concrete corporate governance principles established in and 79 real life corporate charters selected to represent diverse industries from a total dataset of 10k real life corporate charters. Evaluations on state-of-the-art (SOTA) reasoning models confirm the difficulty of the benchmark, with models such as Claude 3.7 Sonnet and GPT-4o achieving 64.5% and 75.2% accuracy respectively. Reasoning agents exhibit superior performance, with agents based on the ReAct and CodeAct frameworks scoring 76.1% and 78.1% respectively, further confirming the advanced legal reasoning capabilities required to score highly on the benchmark. We also conduct an analysis of the types of questions which current reasoning models struggle on, revealing insights into the legal reasoning capabilities of SOTA models.

  PublisherOA PDFDOI

• A transformer model for de novo sequencing of data-independent acquisition mass spectrometry data

  Nature Methods · 2025-07-01 · 9 citations

  article
  PublisherDOI

• POMACS V9, N1, March 2025 Editorial

  Proceedings of the ACM on Measurement and Analysis of Computing Systems · 2025-03-06

  articleOpen access1st authorCorresponding

  The Proceedings of the ACM on Measurement and Analysis of Computing Systems (POMACS) focuses on the measurement and performance evaluation of computer systems and operates in close collaboration with the ACM Special Interest Group SIGMETRICS. All papers in this issue of POMACS will be presented at the ACM SIGMETRICS 2025 conference on June 9-13, 2025, in Stony Brook, New York, USA. The papers in this issue were selected during the fall submission round by the 94 members of the ACM SIGMETRICS 2025 program committee via a rigorous review process. Each paper was conditionally accepted (and shepherded), or allowed to be resubmitted to one of the subsequent two SIGMETRICS deadlines, or rejected (with resubmission allowed only after a year). POMACS is publishing 14 papers out of 113 submissions in this issue, of which 2 had previously received a resubmit decision. One accepted paper opted out of publication in POMACS. Submissions typically receive at least three reviews. While most papers were decided in the online discussion phase, borderline cases were extensively discussed during the online program committee meeting. Based on the indicated track(s), roughly 25% of the submissions were in the Theory track, 25% were in the Measurement &amp; Applied Modeling track, 31% were in the Systems track, and 19% were in the Learning track. Many individuals contributed to the success of this issue of POMACS. We would like to thank the authors, who submitted their best work to SIGMETRICS/POMACS. We would also like to thank the program committee members who provided constructive feedback in their reviews to authors and participated in the online discussions and program committee meeting. Further, we thank the several external reviewers who provided their expert opinions on specific submissions that required additional input. Finally, we are also grateful to the ACM SIGMETRICS Chair, Mor Harchol-Balter, to the SIGMETRICS Organization Committee and to the SIGMETRICS Executive Committee for their ongoing efforts and initiatives for creating an exciting program for ACM SIGMETRICS 2025. We hope that you enjoy reading the papers in this POMACS issue, and we look forward to your continued submissions to the ACM SIGMETRICS conference and the POMACS journal.

  PublisherOA PDFDOI

• Scalable Fingerprinting of Large Language Models

  ArXiv.org · 2025-02-11

  preprintOpen accessSenior author

  Model fingerprinting has emerged as a powerful tool for model owners to identify their shared model given API access. However, to lower false discovery rate, fight fingerprint leakage, and defend against coalitions of model users attempting to bypass detection, we argue that {\em scalability} is critical, i.e., scaling up the number of fingerprints one can embed into a model. Hence, we pose scalability as a crucial requirement for fingerprinting schemes. We experiment with fingerprint design at a scale significantly larger than previously considered, and introduce a new method, dubbed Perinucleus sampling, to generate scalable, persistent, and harmless fingerprints. We demonstrate that this scheme can add 24,576 fingerprints to a Llama-3.1-8B model -- two orders of magnitude more than existing schemes -- without degrading the model's utility. Our inserted fingerprints persist even after supervised fine-tuning on standard post-training data. We further address security risks for fingerprinting, and theoretically and empirically show how a scalable fingerprinting scheme like ours can mitigate these risks. Our code is available at https://github.com/SewoongLab/scalable-fingerprinting-of-llms

  PublisherOA PDFDOI

• Zeroth-Order Optimization Finds Flat Minima

  ArXiv.org · 2025-06-05

  preprintOpen access

  Zeroth-order methods are extensively used in machine learning applications where gradients are infeasible or expensive to compute, such as black-box attacks, reinforcement learning, and language model fine-tuning. Existing optimization theory focuses on convergence to an arbitrary stationary point, but less is known on the implicit regularization that provides a fine-grained characterization on which particular solutions are finally reached. We show that zeroth-order optimization with the standard two-point estimator favors solutions with small trace of Hessian, which is widely used in previous work to distinguish between sharp and flat minima. We further provide convergence rates of zeroth-order optimization to approximate flat minima for convex and sufficiently smooth functions, where flat minima are defined as the minimizers that achieve the smallest trace of Hessian among all optimal solutions. Experiments on binary classification tasks with convex losses and language model fine-tuning support our theoretical findings.

  PublisherOA PDFDOI

• POMACS V9, N2, June 2025 Editorial

  Proceedings of the ACM on Measurement and Analysis of Computing Systems · 2025-05-27

  article1st authorCorresponding

  The Proceedings of the ACM on Measurement and Analysis of Computing Systems (POMACS) focuses on the measurement and performance evaluation of computer systems and operates in close collaboration with the ACM Special Interest Group SIGMETRICS. All papers in this issue of POMACS will be presented at the ACM SIGMETRICS 2025 conference on June 9-13, 2025, in Stony Brook, New York, USA. The papers in this issue were selected during the winter submission round by the 94 members of the ACM SIGMETRICS 2025 program committee via a rigorous review process. Each paper was conditionally accepted (and shepherded), or allowed to be resubmitted to one of the subsequent two SIGMETRICS deadlines, or rejected (with resubmission allowed only after a year).

  PublisherDOI

• Economic and Environmental Performance Improvements Based on S-100 Hydrographic Information

  Sensors and Materials · 2025-02-28 · 1 citations

  articleOpen access
  PublisherOA PDFDOI

Recent grants

• CIF: RI: Small: Information-theoretic measures of dependencies and novel sample-based estimators

  NSF · $450k · 2019–2022

• EAGER: A Graphical Approach for Choice Modeling

  NSF · $88k · 2015–2015

• TWC: Small: Fundamental Limits in Differential Privacy

  NSF · $495k · 2015–2019

• CAREER: Social Computation: Fundamental Limits and Efficient Algorithms

  NSF · $360k · 2016–2019

• CAREER: Social Computation: Fundamental Limits and Efficient Algorithms

  NSF · $394k · 2019–2021

Frequent coauthors

• Pramod Viswanath

  88 shared

• Sreeram Kannan

  49 shared

• Peter Kairouz

  38 shared

• Yihan Jiang

  University of Florida

  31 shared

• Kiran Koshy Thekumparampil

  26 shared

• Ashish Khetan

  25 shared

• Hyeji Kim

  25 shared

• Giulia Fanti

  25 shared

Labs

• Charles H. Sandage Department of AdvertisingPI

Education

• Ph.D., Mass Communications/Public Health

  University of South Carolina

• M.A., Mass Communications

  Sogang University

• B.A., Social Welfare/Mass Communications

  Ewha Womans University

Awards & honors

• Inaugural Teri Thompson Outstanding Article Award for best a…
• JWY Research Award, Department of Advertising, University of…
• Best Research Paper, 2nd place, Global Colloquium, Korea Adv…
• Rainbow Top Research Paper Award, Korea Health Communication…
• Red Raider Public Relations Research Award, International Pu…