Vivek Vrujlal Vekariya, M.Sc.
Technical University of Munich
Informatics 4 - Chair of Software & Systems Engineering (Prof. Pretschner)
Postal address
Postal:
Boltzmannstr. 3
85748 Garching b. München
Place of employment
Informatics 4 - Chair of Software & Systems Engineering (Prof. Pretschner)
Work:
Boltzmannstr. 3(5611)/I
85748 Garching b. München
About Me
| Have you ever wondered why Deep Neural Networks (DNNs), despite their remarkable success in surpassing human-level performance, still face significant hurdles when deployed in safety-critical systems? One possible reason is that the test datasets used to evaluate DNNs may not adequately capture real-world diversity. Ensuring the quality of these test datasets is crucial. |
| To address these challenges, I am currently pursuing my PhD under the guidance of Prof. Pretschner, focusing on exploring the latent space of DNNs and its properties. My research aims to quantify and improve test dataset quality, prioritize test cases, and enhance robustness. |
| Prior to my doctoral studies, I worked at fortiss GmbH for three years, where I gained valuable experience in the field. My journey is driven by the goal of making DNNs more reliable and robust for real-world applications. |
Teaching and Theses Supervision
Ongoing Theses
Unfortunately, I can't supervise any further thesis topics in WS24-25.
| WS 2024-25 | Evaluating Test Adequacy and Coverage Criteria for Deep Neural Networks: A Comprehensive Study (in cooperation with fortiss GmbH) | Assigned | MasterThesis |
| WS 2024-25 | Refining Deep Neural Networks in Radar Systems Through Latent Space Test Data Analysis (in cooperation with Infineon Technologies) | Assigned | Master Thesis |
| SS 2024 | Testcase Prioritization using Latent Space Properties of Deep Neural Networks (DNNs) | Assigned | Bachelor Thesis |
Completed Theses
| SS 2024 | Navigating the Neural Maze: A Strategic Guide for Evaluating and Improving Robustness of Deep Neural Networks | Completed | Guided-Research |
| WS 2023-24 | Investigating Latent Space Testing Boundaries in Deep Learning Models | Completed | Bachelor Thesis |
Courses
| Semester | Title | ||
| WS 2024-25 | Advanced Testing of DL Models: Towards Robust AI | Bachelor/Master Practicum | |
| SS 2024 | Advanced Testing of DL Models: Towards Robust AI | Master Practicum | |
| WS 2023-24 | Advanced Testing of DL Models: Towards Robust AI | Bachelor/Master Practicum | in cooperation with fortiss GmbH |
| WS 2022-23 | Advanced Testing of DL Models: Towards Robust AI | Bachelor/Master Practicum | in cooperation with fortiss GmbH |
Interested in Writing Your Bachelor, Master, or Guided Research Thesis?
| If you are passionate about my research (Testing AI-based Systems) and looking to write your Bachelor, Master, or Guided Research thesis, I would be delighted to hear from you. |
| To apply, please submit the following documents: |
| - Curriculum Vitae (CV) |
| - Cover Letter |
| - Transcripts |
| You can reach me via email for any inquiries or to submit your application. I look forward to potentially working with you and contributing to your academic journey. |
Publications
F. Neugebauer, V. Vekariya, I. Polian, J. P. Hayes; Stochastic Computing as a Defence Against Adversarial Attacks; 2023 53rd Annual IEEE/IFIP International Conference on Dependable Systems and Networks Workshops (DSN-W)