Alexandra (Ola) Zytek

CV - Github - Google Scholar - Project Website - LinkedIn

Machine learning algorithms are becoming increasingly powerful - but how can we extend their benefits to a diverse set of real-world domains? Models continue to be black-boxes that confuse and concern users, and using them can be a difficult and complicated process. 

My research aims to bridge the gap between algorithms and humans through collaborations with end-users and development of software systems and interfaces. Through these methods, ML applications can better support the nuances of real-world domains and users.

I am a PhD student at MIT, working in the Data to AI Lab under the supervision of Kalyan Veeramachaneni.  


Projects

Pyreal 

tutorial - documentationgithub - paper

Python library for low-code generation of ML explanations that are readily understood by users, even those without ML expertise.

Sibyl-API

documentationgithub

Generalizable REST API for readily-understandable explainable ML.

Sibylapp

demo - github - paper 1 - paper 2

Customizable front-end UI for bringing explainable ML into real-world domains.

Explingo

github 

Using LLMs to generate and evaluate more natural, usable ML explanations.


Selected Publications

Google Scholar

Zytek, A., Wang, W. E., Koukoura, S., & Veeramachaneni, K. (2023). Lessons from Usable ML Deployments Applied to Wind Turbine Monitoring. In NeurIPS XAIA.

Zytek, A., Pido, S., Veeramachaneni, K. (2024). LLMs for XAI: Future Directions for Explaining Explanations. To be presented in ACM CHI HCXAI.

Zytek, A., Arnaldo, I., Liu, D., Berti-Equille, L., & Veeramachaneni, K. (2022). The Need for Interpretable Features: Motivation and Taxonomy. In KDD Explorations.

Zytek, A., Liu, D., Vaithianathan, R., & Veeramachaneni, K. (2021). Sibyl: Understanding and Addressing the Usability Challenges of Machine Learning In High-Stakes Decision Making. In IEEE Transactions on Visualization and Computer Graphics (VIS).

Cheng, F., Liu, D., Du, F., Lin, Y., Zytek, A., Li, H., Qu, H. & Veeramachaneni, K. (2021). VBridge: Connecting the Dots Between Features, Explanations, and Data for Healthcare Models. In IEEE Transactions on Visualization and Computer Graphics (VIS). Honorable Mention.

Thesis

Zytek, A. (2021). Towards Usable Machine Learning (S.M. thesis, MIT).


Contact

Email

zyteka at mit dot edu

Office

MIT Stata Center
32 Vassar St., Room 32-D17
Cambridge, MA 02139