Mirazul Haque

ABOUT ME

The University of Texas at Dallas

Hello, my name is Mirazul Haque.

I am currently pursuing PhD in the field of Computer Science at University of Texas at Dallas. I am working under the supervision of Dr. Wei Yang and my research interest includes AI Security and Software Engineering. In 2015, I have received my bachelors degree in Electronics and Communications Engineering from Heritage Institute of Technology, Kolkata. Also, I am a recipient of Siemens FutureMakers Fellowship for year 2021-22.

Feel free to connect with me on LinkedIn or even shoot me an email.

RESEARCH PROJECTS

With the increase in the number of layers and parameters in neural networks, energy consumption of neural networks has become a great concern to society, especially to users of handheld or embedded devices. In this project, we investigate the robustness of neural networks against energy-oriented attacks. Specifically, we propose ILFO (Intermediate Output Based Loss Function Optimization) attack against a type of energy-saving neural networks, Adaptive Neural Networks (AdNN).
An AdNN can dynamically deactivate part of its model based on the need of the inputs to decrease energy consumption. ILFO (Intermediate Output Based Loss Function Optimization) leverage intermediate output as a proxy to infer the relation between input and its corresponding energy consumption.
ILFO has shown an increase up to 100 % of the remaining FLOPs (floating-point operations per second) count of AdNNs with minimum noise added to input images. To our knowledge, this is the first attempt to attack the energy consumption of a DNN.

You may read more about it here.

Because of the increasing accuracy of Deep Neural Networks (DNNs) on different tasks, a lot of real times systems are utilizing DNNs. These DNNs are vulnerable to adversarial perturbations and corruptions. Specifically, natural corruptions like fog, blur, contrast etc can affect the prediction of DNN in an autonomous vehicle. In real time, these corruptions are needed to be detected and also the corrupted inputs are needed to be de-noised to be predicted correctly. You may read more about it here.

CONFERENCE PUBLICATIONS

ILFO: Adversarial Attacks on Adaptive Neural Networks

Authors: Mirazul Haque, Anki Chauhan, Cong Liu, Wei Yang

In The Conference on Computer Vision and Pattern Recognition (CVPR) 2020

You may read it here

EREBA: Black-box Energy Testing of Adaptive Neural Networks

Authors: Mirazul Haque*, Yaswanth Yadlapalli*, Wei Yang, Cong Liu

In International Conference on Software Engineering (ICSE) 2022

You may read it here

CorrGAN:Input Transformation Technique Against Natural Corruptions

Authors: Mirazul Haque, Christof Budnik, Wei Yang

In Workshop at The Conference on Computer Vision and Pattern Recognition (CVPR-W) 2022

You may read it here

NICGSlowDown: Evaluating the Efficiency Robustness of Neural Image Caption Generation Models

Authors: Simin Chen, Zihe Song, Mirazul Haque, Cong Liu, Wei Yang

In The Conference on Computer Vision and Pattern Recognition (CVPR) 2022

You may read it here

NMTSloth: Understanding and Testing Efficiency Degradation of Neural Machine Translation Systems

Authors: Simin Chen, Cong Liu, Mirazul Haque, Zihe Song, Wei Yang

In ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE) 2022

DeepPerform: An Efficient Performance Testing Framework for Adaptive Neural Networks

Authors: Simin Chen, Mirazul Haque, Cong Liu, Wei Yang

In IEEE/ACM International Conference on Automated Software Engineering (ASE) 2022

TestAug: A framework for Augmenting Capability-based NLP Tests

Authors: Guanqun Yang, Mirazul Haque, Qiaochu Song, Wei Yang, Xueqing Liu

In International Conference on Computational Linguistics (COLING) 2022

MIRAZUL HAQUE

WELCOME TO MY WEBSITE

ABOUT ME

RESEARCH PROJECTS

CONFERENCE PUBLICATIONS

ILFO: Adversarial Attacks on Adaptive Neural Networks

EREBA: Black-box Energy Testing of Adaptive Neural Networks

CorrGAN:Input Transformation Technique Against Natural Corruptions

NICGSlowDown: Evaluating the Efficiency Robustness of Neural Image Caption Generation Models

NMTSloth: Understanding and Testing Efficiency Degradation of Neural Machine Translation Systems

DeepPerform: An Efficient Performance Testing Framework for Adaptive Neural Networks

TestAug: A framework for Augmenting Capability-based NLP Tests

EXPERIENCE

WORK

Siemens

Siemens

Apple Inc.

Futurewei Technologies

Infosys Technologies

TEACHING ASSISTANCE

Database Design (CS 6360)

Operating Systems (CS 4348)

Education

The University of Texas at Dallas

The University of Texas at Dallas

CURRICULAM VITAE