AKSHAY KULKARNI
  • Home
  • Research Experience
  • Projects
    • Post-hoc Gen-CBM (CVPR25)
    • IG-Defense (ECCV24)
    • Mixup-SFDA (ICML22)
    • SFDA-Seg (ICCV21)
    • Sticker-SFDA (ECCV22)
    • SPA-UniDA (NeurIPS22)
    • AST-OCDASeg (AAAI22)
    • Autonomous Driving Platform
    • Semantic Segmentation using PyTorch Lightning
    • Human Activity Recognition using IMU
    • Real-Time Stair Detection
    • Mini-Projects >
      • ESP32 & IMU interfacing with ROS
      • Implementation of DL Models
      • Hand Written Symbol Recognition
      • Wizards' Chess
      • Snake Gaits Implementation
      • Maze Solver
      • Sudoku Solver
    • Simulations >
      • Offboard Control of Pixhawk
      • Turtlesim Experiments
      • 3DOF Robotic Arm
  • Notes

AKSHAY KULKARNI

I am a first-year PhD student at the Univ. of California, San Diego (UCSD) advised by Prof. Tsui-Wei (Lily) Weng. I recently completed my MS CS thesis at UCSD, working with Prof. Weng on Interpretability-Guided Adversarial Robustness. My major research interests are in Interpretability, Adversarial Robustness, and Domain Adaptation for Deep Neural Networks.

During Summer 2023, I worked as a Research Intern at Sony R&D Center, Tokyo on Labeled Dataset Generation using Generative AI models.

Before starting MS, I was a Project Assistant at Video Analytics Lab, IISc, Bangalore with Prof. R. Venkatesh Babu and his Ph.D. student, Jogendra Nath Kundu.
My work there was focused on empirical and theoretical aspects of Domain Adaptation.

I completed B. Tech. (Electrical & Electronics Engg.) (2016-20) from Visvesvaraya National Institute of Technology, Nagpur, where
I was a core coordinator at IvLabs (Robotics Club).
Picture
Email | GitHub | Google Scholar | Stack Overflow
News
  • Apr. 2025 - Honored to receive UCSD GPSA Travel Grant and CVPR Broadening Participation Travel Grant for CVPR 2025
  • Apr. 2025 - Recognized as one of the Best Reviewers for AISTATS 2025
  • Mar. 2025 - Serving as a reviewer for ICML 2025 and ICCV 2025
  • Feb. 2025 - Our paper on Concept Bottleneck Generative Models is accepted at CVPR 2025
  • Jan. 2025 - Our paper on Interpreting Neurons is accepted by TMLR
  • Sept. 2024 - Recognized as an Outstanding Reviewer for ECCV 2024
  • July 2024 - Our paper on Interpretability-Guided Adversarial Robustness is accepted at ECCV 2024
  • ​Older news
Selected Publications    *equal contribution
See complete list or Google Scholar page

Picture
Interpretable Generative Models through Post-hoc Concept Bottlenecks ​
Akshay Kulkarni, Ge Yan*, Chung-En Sun*, Tuomas Oikarinen, and Tsui-Wei Weng
CVPR 2025
Picture
Interpretability-Guided Test-Time Adversarial Defense
Akshay Kulkarni and Tsui-Wei Weng
ECCV 2024
Picture
Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation
J. N. Kundu*, Akshay Kulkarni*, A. Singh, Varun Jampani, and R. Venkatesh Babu
ICCV 2021
Picture
Balancing Discriminability and Transferability for Source-Free Domain Adaptation
J. N. Kundu*, A. Kulkarni*, S. Bhambri*, D. Mehta, S. Kulkarni, V. Jampani, R. V. Babu
ICML 2022
Picture
Concurrent Subsidiary Supervision for Unsup. Source-Free Domain Adaptation
J. N. Kundu*, S. Bhambri*, Akshay Kulkarni*, H. Sarkar, V. Jampani, and R. V. Babu
ECCV 2022
Picture
Subsidiary Prototype Alignment for Universal Domain Adaptation
J. N. Kundu*, S. Bhambri*, Akshay Kulkarni*, H. Sarkar, V. Jampani, and R. V. Babu
NeurIPS 2022
Picture
Interpreting Neurons in Deep Vision Networks with Language Models
N. Bai*, R. A. Iyer*, Tuomas Oikarinen, Akshay Kulkarni, and Tsui-Wei Weng
TMLR 2025
Picture
Domain-Specificity Inducing Transformers for Source-Free Domain Adaptation
S. Sanyal*, A. Asokan*, S. Bhambri*, Akshay Kulkarni, J. N. Kundu, and R. V. Babu
ICCV 2023
Picture
Amplitude Spectrum Transformation for Open Compound Domain Adaptation
J. N. Kundu*, Akshay Kulkarni*, S. Bhambri*, Varun Jampani, and R. Venkatesh Babu
AAAI 2022
Picture
Aligning Non-Causal Factors for Source-Free Domain Adaptation
S. Sanyal*, A. Asokan*, S. Bhambri*, Pradyumna, A. Kulkarni, J. N. Kundu, R. V. Babu
WACV 2024
Academic Service
  • Reviewer:
    • CV conferences: WACV22-25, ECCV22/24, CVPR23-24, ICCV23/25, ACCV24
    • ML conferences: ICML22/24-25, NeurIPS22-24, ICLR24, ACML22, CoLLAs23-24, AISTATS25
    • Journals: IEEE TPAMI, IEEE TIP
  • Workshop Papers Reviewer: LatinXCV (CVPR22-23, ECCV22), Shift Happens (ICML22), MLRC22, WiCV (CVPR23), SCIS (ICML23), XAIA (NeurIPS23), CRL (NeurIPS23)
  • Teaching Assistant: DSC291 (Trustworthy Machine Learning, Spring '23, UCSD), DSC210 (Numerical Linear Algebra, Fall '23, UCSD), DSC291 (Trustworthy Machine Learning, Spring '24, UCSD)
Awards
  • Outstanding Reviewer: ECCV 2024, AISTATS 2025
  • Travel Grants: UCSD GPSA Spring 2025 and CVPR Broadening Participation for attending CVPR 2025
Powered by Create your own unique website with customizable templates.
  • Home
  • Research Experience
  • Projects
    • Post-hoc Gen-CBM (CVPR25)
    • IG-Defense (ECCV24)
    • Mixup-SFDA (ICML22)
    • SFDA-Seg (ICCV21)
    • Sticker-SFDA (ECCV22)
    • SPA-UniDA (NeurIPS22)
    • AST-OCDASeg (AAAI22)
    • Autonomous Driving Platform
    • Semantic Segmentation using PyTorch Lightning
    • Human Activity Recognition using IMU
    • Real-Time Stair Detection
    • Mini-Projects >
      • ESP32 & IMU interfacing with ROS
      • Implementation of DL Models
      • Hand Written Symbol Recognition
      • Wizards' Chess
      • Snake Gaits Implementation
      • Maze Solver
      • Sudoku Solver
    • Simulations >
      • Offboard Control of Pixhawk
      • Turtlesim Experiments
      • 3DOF Robotic Arm
  • Notes