Mehrnaz Mofakhami
I am a Visiting Researcher at ServiceNow
Research and
a master's student at Université de
Montréal and Mila, where I am supervised by Prof. Gauthier Gidel and Prof. Ioannis Mitliagkas. Prior to coming to
Montreal, I got my bachelor's degree at Sharif University
of Technology in Computer Science.
I am interested in the fundamentals of machine learning, inference optimization
and the development of efficient algorithms. I get excited about exploring how AI
can be leveraged effectively and ethically to address real-world challenges and improve decision-making
processes across various domains. My master's research revolved around Performative Prediction, a framework
for learning models that actively shape the environment they interact with through their predictions. The
primary objective of my research was to push the boundaries of this framework,
exploring and identifying new domains where its application can be extended.
You can reach me at: mehrnaz (dot) mofakhami [at] mila.quebec
 
 
 
 
|
|
Recent News
- August 2024: I'm excited to announce that I'm serving as the Mentorship and Networking Program Chair for the Women in Machine Learning (WiML) workshop at NeurIPS 2024.
- July 2024: I presented my work on "Performance Control in Early Exiting to Deploy Large Models at the Same Cost of Smaller Ones" at the ES-FoMO workshop at ICML 2024 in Vienna.
- June 2024: I served as a reviewer for the ES-FoMO workshop during ICML 2024, and previously at AISTATS 2024.
- May 2024: I am very grateful to have received the AI Scholarship for Excellence in Research from the University of Montreal's Graduate and Postdoctoral Studies.
- March 2024: Happy to have received a full travel grant to attend the Cornell, Maryland, Max Planck Pre-doctoral Research School in Computer Science (CMMRS) in Summer 2024.
- November 2023: I started an internship with ServiceNow Research at the Multimodal Foundation Models team!
- Fall 2023: I got the merit-based Excellence Scholarship from DIRO (Department of Computer Science and Operational Research) at the University of Montreal for the third time in a row!
- June 2023: I presented an introduction to the Performative Prediction framework and my research on this topic at a Google DeepMind Montreal Tea Talk.
- September 2022: I joined the Space Committee at Mila to guide decisions surrounding the management of the space for students and profs.
|
Performance Control in Early Exiting to Deploy Large Models at the Same Cost of Smaller Ones
Mehrnaz Mofakhami,
Reza Bayat,
Ioannis Mitliagkas,
João Monteiro*,
Valentina Zantedeschi*
Efficient Systems for Foundation Models Workshop, ICML 2024
paper
We provided a method for performance control in early exiting and showed that larger models coupled
with early exiting using this method can achieve lower prediction errors for the same computational budget as smaller models.
|
Performative Prediction with Neural Networks
Mehrnaz Mofakhami,
Ioannis Mitliagkas,
Gauthier Gidel
AISTATS 2023
paper /
video
We analyzed the performative prediction framework in the presence of Neural Networks with
non-convex loss function using functional analysis.
A preliminary version of this work was presented at NeurIPS 2022 workshop on Distribution
Shifts.
|
Reproduction: Adversarial Example Games
Adversarial Machine Learning course
Report /
code
In this project, I reproduced the simple setup described in the Adversarial Example Games paper on a binary classification task with logistic regression.
|
Reproduction: Tracking the World State with Recurrent Entity Networks
Mehrnaz Mofakhami,
AmirHossein Yavari
EEML Summer School 2021
- Best poster award
Report /
code
|
|