A Novel Metric for Measuring the Robustness of Large Language Models in Non-adversarial Scenarios
- Samuel Ackerman
- Ella Rabinovich
- et al.
- 2024
- EMNLP 2024
I earned my PhD in statistics in 2018 from Temple University in Philadelphia. My research there involved using Bayesian computational particle filtering algorithms to model the movement patterns of sharks based on inferring their behavioral state. I am a proficient R programmer and have designed two software packages (mapStats and animalEKF) for users. At IBM, I primarily work on on integrating statistical inference and data science tools into applications, particularly in changepoint detection (e.g., statistical guarantees on correctness, such as using Bayes Factors), on the FreaAI system, and in analysis and testing of transformer networks.
I have a blog through IBM (http://www.research.ibm.com/haifa/dept/vst/ML-QA.shtml) where I address broad-interest (and perhaps under-addressed) issues in statistics and data science, such as best practices in statistical inference.