Bogdan Kulynych
/bogh-duhn koo-lin-ich/
Ph.D. candidate in Computer Science at EPFL. B.Sc. from Kyiv Mohyla Academy in Ukraine. Formerly a research fellow at Harvard SEAS, intern at Google and CERN. I study privacy, security, reliability, and broader societal harms of algorithmic systems.
Twitter: @hiddenmarkov. Email: [first name] at kulyny.ch
Advocacy: No peace without justice. Many people believe that pursuing "peace at all costs" is the right course of action in Ukraine. While appearing moral, halting military action at the current stage means that the mass killings, summary executions, rape and torture, deportations, arbitrary violence and detention, looting of people's belongings and cultural heritage in the occupied territories will continue with impunity. Not only "peace at all costs" is a call for subjugation in disguise, but it cannot stop even the military action in the long term. After a pause, the attack is bound to resume with fresh forces unless the mainstream imperialist chauvinist ideology loses its cultural and political domination in Russia. If you have political representation in one of the countries that are capable of providing assistance to Ukraine, please call for aid and sanctions that would help Ukraine to achieve not only peace but justice — at the very least, the liberation of the occupied territories — for as long as Ukrainians are willing to fight for it.
Host scholars and students. Consider hosting Ukrainian scholars and students at your institutions through, e.g., Science for Ukraine.
Platform Ukrainian voices. If you hold events or curate publications on Ukraine, make sure to invite Ukrainians and experts on Ukraine, not Poland or Russia. If you are looking for Ukrainian academics who could comment on the war taking into account the technological aspects, I have compiled a list for you.
Donate to vetted local organizations on the ground. The majority of humanitarian work in Ukraine is done by local organizations, yet they receive only 0.003% of all humanitarian contributions. Reluctance to donate to unfamiliar organizations is understandable, but I encourage you to try this curated list of local organizations: standforukraine.com See more resources here.
Latest and Upcoming Talks
-
Arbitrary Decisions are a Hidden Cost of Differentially Private Learning
European Workshop on Algorithmic Fairness (EWAF). Jun 07, 2023
-
The Good and the Bad: Side Effects of Differentially Private Learning
ETH AI Center. Mar 14, 2023
-
What You See is What You Get: Principled Deep Learning via Distributional Generalizaton
NeurIPS. Poster presentation. Nov 29, 2022
[details] -
What You See is What You Get: Principled Deep Learning via Distributional Generalizaton
MIT. Talk at Marzyeh Ghassemi's lab group meeting. Nov 18, 2022
-
Disparate Vulnerability to Membership Inference Attacks
Northeastern University. Talk at Alina Oprea's group meeting. Nov 16, 2022
Academic Publications and Texts
Prediction without Preclusion: Recourse Verification with Reachable Sets
A. Kothari, B. Kulynych, T. Weng, B. Ustun. Coming up soon, 2023
Arbitrary Decisions are a Hidden Cost of Differentially Private Training
B. Kulynych, H. Hsu, C. Troncoso, F. Calmon. To appear in FAccT, 2023
[arxiv]
Adversarial Robustness for Tabular Data through Cost and Utility Awareness
K. Kireev*, B. Kulynych*, C. Troncoso. NDSS, 2023
[arxiv]
[twitter thread]
What You See is What You Get: Principled Deep Learning via Distributional Generalization
B. Kulynych*, Yao-Yuan Yang*, Y. Yu, J. Blasiok, P. Nakkiran. NeurIPS, 2022
[arxiv]
[video]
[twitter thread]
Disparate Vulnerability to Membership Inference Attacks
B. Kulynych, M. Yaghini, G. Cherubin, M. Veale, C. Troncoso. PETS, 2022
[arxiv]
[twitter thread]
Exploring Data Pipelines through the Process Lens: a Reference Model for Computer Vision
A. Balayn, B. Kulynych, S. Gürses. CVPR "Beyond Fair Computer Vision" Workshop, 2021
[arxiv]
Adversarial for Good? How the Adversarial ML Community's Values Impede Socially Beneficial Uses of Attacks
K. Albert*, M. Delano*, B. Kulynych*, R. Shankar Siva Kumar*. ICML "A Blessing in Disguise: The Prospects and Perils of Adversarial Machine Learning" Workshop, 2021
[arxiv]
[video]
[twitter thread]
Protective Optimization Technologies
B. Overdorf*, B. Kulynych*, C. Troncoso, S. Gurses. FAccT, 2020
[arxiv]
[video]
[twitter thread]
Questioning the Assumptions Behind Fairness Solutions
B. Overdorf*, B. Kulynych*, E. Balsa, C. Troncoso, S. Gurses. NeurIPS Critiquing and Correcting Trends in ML, 2018
[arxiv]
Evading Classifiers in Discrete Domains with Provable Optimality Guarantees
B. Kulynych, J. Hayes, N. Samarin, C. Troncoso. NeurIPS Workshop on Security and Privacy in ML, 2018
[arxiv]
zksk: A Library for Composable Zero-Knowledge Proofs
W. Lueks, B. Kulynych, J. Fasquelle, S. Le Bail-Collet, C. Troncoso. WPES, 2019
[arxiv]
ClaimChain: Improving the Security and Privacy of In-band Key Distribution for Messaging
B. Kulynych*, M. Isaakidis*, Wouter Lueks, George Danezis, Carmela Troncoso. WPES, 2018
[arxiv]
Media Coverage and Comments
- Twitter algorithmic bias challenge: The Guardian, Wired, The Verge, CNET
- Premise app safety concerns in the beginning of 2022 Russia's invasion of Ukraine: The Wall Street Journal
Latest Academic Service
- FAccT 2022 – 2023. Program committee member
- ICML 2023. Reviewer
- The Web Conference 2023 FATE. Program committee member
- NeurIPS 2022. Reviewer
- NeurIPS 2022 – 2021. Ethics reviewer
- LSE Justice, Equity, and Technology Table (JETT). Advisory board member
- Participatory Approaches to Machine Learning Workshop at ICML 2020. Co-organizer
- Workshop reviewing: ICLR 2023 Trustworthy ML, FAccT CRAFT Track 2021, NeurIPS 2021 Privacy in Machine Learning Workshop, NeurIPS 2021 I Can’t Believe It’s Not Better! (ICBINB), ICLR 2020 Towards Trustworthy ML
