Vishnu Asutosh Dasu
vdasu at psu dot edu

I am a PhD student in computer science (PhD CSE) at the Pennsylvania State University. I am advised by Prof. Gang Tan and work on developing secure and trustworthy software and machine learning systems. Currently, I am working on trustworthy code generation LLMs and fairness of LLMs. For my master's thesis, I worked on mitigating unfairness issues in trained neural networks. I also enjoy collaborating with Prof. Shagufta Mehnaz on privacy of LLMs and security of multimodal LLMs. In general, I am interested in topics in the intersection security, privacy, and machine learning. In another life, I worked on hardware security and cryptography.

CV  /  Research Summary  /  Google Scholar  /  LinkedIn  /  Twitter  /  GitHub

My Erdős number is 4: Paul Erdős → Israel Koren → Francesco Regazzoni → Takanori Isobe → Vishnu Asutosh Dasu

I would love to collaborate with like-minded researchers and am open to opportunities. Please reach out to me if you're interested!

(* represents equal contribution)

We develop a novel variant of the Private Set Intersection (PSI) protocol that securely removes all pairwise duplicates among datasets held by two more users in federated learning. Our protocl preserves user privacy and does not reveal any information about the datasets to the users and server. The deduplication protocol results in significant improvement to LLM perplexity and reduces GPU training time.

We develop randomized algorithms to mitigate unfairness in deep learning by dropping a subset of neurons from a trained neural network during at inference time. Our results show that a subset of neurons disparately contributes to unfairness and dropping them out during inference can improve fairness by up to 69%.

We introduce two novel privacy leakage attacks against federated language models. First, we show that intermediate model snapshots can leak more sensitive data than the final trained model. Second, we show that tampering with a model's selective weights responsible for memorizing sensitive data can aggravate privacy leakage. Our best-performing method outperforms existing attacks with stronger adversary assumptions.

EvoquerBot is a multimedia chatbot developed for the TaskBot challenge, aimed at assisting users with cooking and DIY tasks in a single session. The bot addresses challenges like short development time, data quality, multimedia responses, and tailored conversation flow using agile classifier development, data augmentation, multimedia response design, and domain-specific dialogue state machines, ultimately improving user experience through superior task recommendations.

We show new machine learning differential distinguishers for unkeyed and round-reduced versions of SPECK-32, SPECK-128, ASCON, SIMECK-32, SIMECK-64, and SKINNY-128. Our comprehensive experiments utilize neural networks and support vector machines in various settings and numerous input difference tuples.

We propose PROV-FL, a secure and private federated learning protocol. PROV-FL utilizes homomorphic encryption and differential privacy to provide strong privacy guarantees. It is resilient to user dropouts/joins, supports verifiable aggregation, and requires only a single round of communication without a full-trusted third party.

We propose an end-to-end solution to perform SCA on stream ciphers by combining automated tools such as ML, MILP, and SMT. We demonstrate its efficacy by taking electromagnetic traces from a 32-bit software platform and performing SCA on the TRIVIUM stream cipher.

We develop a method to extend the Boyar-Peralta's algorithm to use XOR3 gates, add XOR3 gates to existing XOR2 implementations, and show several SOTA results on the linear layers of block ciphers using different logic libraries.

We provide new insights for different notions of XOR count, develop methods to find sequential XOR count implementations using SMT and MILP, and present new results using XOR3 gates on the AES MixColumn matrix.

We develop a framework that extends LIGHTER to add support for generating optimized implementations of 4x4 SBoxes using reversible logic libraries.

We reproduce the results and validate the claims presented in GANSpace: Discovering Interpretable GAN Controls.

1. TCS Citation Award (3-time Recipient): Received the TCS Citation Award and appreciation from the Chief Technical Officer and Head of TCS Research thrice for performance and outstanding contribution to the organization.
2. Scholarship: Received a scholarship to attend the Winter School on Responsible AI in The Dead Sea, Israel.
3. Best Project Award: Received the Best Project Award among 13 teams during the Fifth Summer School on Computer Vision, Graphics and Image Processing, Indian Statistical Institute (ISI) Kolkata.
4. IGVC: Placed 2nd in the Interoperability Profiles Challenge and 9th overall at Intelligent Ground Vehicle Competition (IGVC) 2018 among 26 teams. Second-best among all teams from India.
5. ACM ICPC Regionals: Represented Manipal Institute of Technology, Manipal at the 2017 ACM ICPC Asia Regional Contest.
6. DAGsHub Award: Received a $500 award from DAGsHub for successfully reproducing GANSpace: Discovering Interpretable GAN Controls and completing the ML Reproducibility Challenge Spring 2021.

• External Reviewer for Networking and Distributed Security Symposium (NDSS), 2025
• Organizer for Penn State Security Reading Group, 2024
• Reviewer for ReScience, 2022 -

• Secure Code Generation using LLMs at Penn State Security Reading Group (slides)

We develop a technique to determine the 3D coordinates of a human from a live video feed using a camera with a single lens (monocamera setup).

I enjoy powerlifitng, playing the guitar, and reading about history, theology, and philosophy.