Research Assistant and PhD student at University of California, San Diego, CSE department

La Jolla, CA

f.mireshghallah@gmail.com


Skills

Documentation (Office Word, Visio, PowerPoint, Excel, LaTex)

90%

C, C++, Python Programming

90%

PyTorch

80%

Verilog HDL

80%

Linux (Ubuntu) and Bash Scripting

80%

MATLAB Scripting

75%

PSPICE & HSPICE

65%

Simulink, Stateflow Toolbox (MATLAB), TensorFlow

40%

Languages

Persian

English

French



FatemehSadat Mireshghallah

"Deep in the sea are riches beyond compare. But if you seek safety, it is on the shore."
-Sa'adi, Iranian Poet

Research Interests

  • Privacy for ML
  • Architectural Support for Machine Learning
  • Computer Architecture
  • Google Scholar
    You can find code for my projects at my GitHub.

    News

  • December 2019: I was chosen as an NCWIT (National Center for Women and IT) collegiate award finalist!

  • December 2019: Join us in Vancouver for the WiMLDS [NeurIPS Special] Talks + Panel Discussion where I'll be giving a talk on Privacy in Mahcine Learning! You can find my slides for this talk here.

  • November 2019: Our paper DeepMist: Towards Protecting Privacy by Learning Noise Distributions for Partial Neural Inference on the Edge got into 25th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS 20) with less than 18% acceptance rate!

  • October 2019: Our paper Shredder got into NeurIPS19's Privacy in ML workshop!

  • June 2019: I am joining Western Digital's research department as a RAMP Next Generation Platform Technologies Intern.

  • June 2019: Our paper Shredder got into ICML's SPML workshop, let's meet up if you are attending ICML19!

  • April 2019: I am attending ASPLOS19, let me know if you are there!
  • Keep up with me on Twitter for more news!


    Publications

  • Shredder: Learning Noise Distributions to Protect Inference Privacy, Under review
  • Energy-Efficient Permanent Fault Tolerance in Hard Real-Time Systems, IEEE Transactions on Computers, March 2019
  • ReLeQ: An Automatic Reinforcement Learning Approach for Deep Quantization of Neural Networks, NeurIPS ML for systems workshop, December 2018

  • Research Experience at ACT lab, University of California, San Diego
    September 2018-Present
    To be updated

    Research Supervisor: Prof. Hadi Esmaeilzadeh


    Research Experience at HPCAN, Sharif University of Technology
    January 2018-May 2018
    My second project was "Energy-Efficient Representation of Metadata Tables of Hardware Pre-fetchers". In this work, we propose a general architecture for prefetchers that reduces the area overhead by eliminating useless metadata and drastically diminish the energy overhead by practicing selective search instead of parallel associative search. Analytical estimates show roughly 30% / 70% reduction in area/energy while preserving performance. I used Verilog in this project to implement different structures (VLDP, SMS and our own design) and to find the area and power consumption of these structures.

    Spring and Summer of 2017

    "Fault Tolerance in Real-Time Systems"
    Real-time systems employ redundancy for reliability which causes huge energy consumption. Many prior energy-reduction techniques result in early failure due to hard faults. In this work, we proposed a method that minimizes energy consumption while tolerating hard faults and incurs only 1% energy overhead in comparison to the state-of-the-art method that does not tolerate hard faults. I used C++ and MATLAB to run simulations and solve optimization problems in this work.


    Volunteer TA Experiences, UC San Diego
    Winter and Fall 2019

  • TA of Accelerator Design for Deep Learning, Graduate Leve, Instructor: Dr. Hadi Esmaeilzadeh
  • Volunteer TA Experiences, Sharif University of Technology
    Fall 2017

  • Head TA of Digital Electronics Course, Instructor: Dr. Siavash Bayat
  • Head TA of Probability and Statistics Course, Fall of 2017, Instructor: Dr. Mohammed Gharib
  • Spring 2017

  • TA of Computer Architecture Course, Instructor: Prof. Hamid Sarbazi-Azad
  • TA of Signals and systems, Spring of 2017, Instructor: Dr. Siavash Bayat
  • Head TA of Probability and Statistics Course, Instructor: Dr. Mohammed Gharib
  • Fall 2016

  • TA of Advanced Programming course, Instructor : Mr. Omid Jafarinezhad
  • Head TA of Numerical Methods course, Instructor: Dr.Mohammed Gharib

  • About Me!

    I like mysteries and puzzles a lot, that is why I enjoy watching police mystery TV series or medical mysteries. I also do clothes designing in my free time, a refreshing activity that helps me focus and clear my mind. I sometimes go jogging after my classes, to get some air and think about ideas for my research. I usually have helpful ideas when I am running or walking and have solved a lot of the problems in my projects while exercising!


    Hobbies

    Swimming, Jogging, Clothes Designing, Etymology and language studies


    Curriculum Vitae