6 papers
Recalling Too Well: Sycophancy Evaluation and Mitigation in Memory-Augmented Models
Shelly Bensal, Axel Magnuson, Aparna Balagopalan +1
Persistent memory systems promise to make LLMs more helpful by storing user beliefs over time. We show they also make models less correct by amplifying sycophancy, wherein models p…
The Price of Agreement: Measuring LLM Sycophancy in Agentic Financial Applications
Zhenyu Zhao, Aparna Balagopalan, Adi Agrawal +3
Given the increased use of LLMs in financial systems today, it becomes important to evaluate the safety and robustness of such systems. One failure mode that LLMs frequently displa…
Bias Delayed is Bias Denied? Assessing the Effect of Reporting Delays on Disparity Assessments
Jennah Gosciak, Aparna Balagopalan, Derek Ouyang +3
Conducting disparity assessments at regular time intervals is critical for surfacing potential biases in decision-making and improving outcomes across demographic groups. Because d…
LEMoN: Label Error Detection using Multimodal Neighbors
Haoran Zhang, Aparna Balagopalan, Nassim Oufattole +4
Large repositories of image-caption pairs are essential for the development of vision-language models. However, these datasets are often extracted from noisy data scraped from the…
What's in a Query: Polarity-Aware Distribution-Based Fair Ranking
Aparna Balagopalan, Kai Wang, Olawale Salaudeen +2
Machine learning-driven rankings, where individuals (or items) are ranked in response to a query, mediate search exposure or attention in a variety of safety-critical settings. Thu…
Event-Based Contrastive Learning for Medical Time Series
Hyewon Jeong, Nassim Oufattole, Matthew Mcdermott +4
In clinical practice, one often needs to identify whether a patient is at high risk of adverse outcomes after some key medical event. For example, quantifying the risk of adverse o…