Piotr (Peter) Mardziel

About me

I do security in AI at realmlabs. Woof woof.

NOTE TO EDITORS: I review articles only in open access journals hosted by non-profit organizations.

Unsorted

• How much pi does gpt know?
• Drift Metrics: How to Select the Right Metric to Analyze Drift
• Drift in Machine Learning: Why it's hard and what to do about it
• trulens: python library for explaining deep models (trulens_explain) and evaluating large language model applications (trulens_eval).
• pyexprlens: lenses from expressions in python.

Courses

• Trustworthy Machine Learning (Stanford CS 329T Spring 2021)
• Security and Fairness of Deep Learning (CMU 18-739 Spring 2020)
• Foundations of Privacy (CMU 18-734 Fall 2018)

Research

topics.{html,txt}; papers.{html,txt}; service.{html,txt}; google scholar; dblp; conflicts.txt; bio.txt

ML Security, Privacy, Fairness

• 2024 PPAI De-amplifying Bias from Differential Privacy in Language Model Fine-tuning.
• 2020 Logic, Language, and Security: Essays Dedicated to Andre Scedrov on the Occasion of His 65th Birthday Gender Bias in Neural Natural Language Processing.
• 2020 NeurIPS Smoothed Geometry for Robust Attribution.
• 2020 AAAI An Information-Theoretic Quantification of Discrimination with Exempt Features.
• 2017 CCS Use Privacy in Data-Driven Systems: Theory and Experiments with Machine Learnt Programs.

ML Interpretability

• 2020 ACL Influence Paths for Characterizing Subject-Verb Number Agreement in LSTM Language Models.
• 2020 CVPR Workshop on Fair, Data Efficient and Trusted Computer Vision Interpreting Interpretations: Organizing Attribution Methods by Criteria.
• 2020 CVPR Workshop on Fair, Data Efficient and Trusted Computer Vision Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks.

Security and Privacy

• 2020 Foundations of Probabilistic Programming, Cambridge University Press Probabilistic Abstract Interpretation: Sound Inference and Application to Privacy.
• 2020 ACM Transactions on Privacy and Security (TOPS) Build It, Break It, Fix It: Contesting Secure Development.
• 2017 POST Quantifying vulnerability of secret generation using hyper-distributions.
• 2016 CCS Build It, Break It, Fix It: Contesting Secure Development.
• 2016 FCS Towards the Quantification of Strategy Leakage.
• 2015 CSF Picking vs. Guessing Secrets: A Game-Theoretic Analysis.
• 2015 CSET Build It Break It: Measuring and Comparing Development Security.
• 2014 S&P/Oakland Quantifying Information Flow for Dynamic Secrets.
• 2014 FCS Adversary Gain vs. Defender Loss in Quantified Information Flow.
• 2013 JCS Dynamic Enforcement of Knowledge-based Security Policies using Probabilistic Abstract Interpretation.
• 2013 PLAS Knowledge Inference for Optimizing Secure Multi-party Computation.
• 2012 PLAS Knowledge-Oriented Secure Multiparty Computation.
• 2011 CSF Dynamic Enforcement of Knowledge-based Security Policies.

Static (and Probabilistic) Analysis

• 2020 Foundations of Probabilistic Programming, Cambridge University Press Probabilistic Abstract Interpretation: Sound Inference and Application to Privacy.
• 2018 ESOP Evaluating Design Tradeoffs in Numeric Static Analysis for Java.
• 2013 JCS Dynamic Enforcement of Knowledge-based Security Policies using Probabilistic Abstract Interpretation.
• 2012 PLAS Knowledge-Oriented Secure Multiparty Computation.
• 2011 CSF Dynamic Enforcement of Knowledge-based Security Policies.