Publications

HyCache: Hybrid Caching for Accelerating Input Preprocessing Pipelines in DNN Training

Author(s): K. Vinayak, S. Pandey, M. Annavaram, A. Basu

The 2025 USENIX Annual Technical Conference, Apr 2025.

Paper Link

On Using Arabic Language Dialects in Recommendation Systems

Author(s): A. Alshabanah, M. Annavaram

The Findings of the Nations of the Americas Chapter of the Association for Computational Linguistics, Apr 2025.

Paper Link

KVPR: Efficient LLM Inference with I/O-Aware Partial KV Cache Recomputation

Author(s): C. Jiang, L. Gao, H. Entezari Zarch, M. Annavaram

AAAI Workshop on Scalable and Efficient Artificial Intelligence Systems, Mar 2025.

Paper Link

MPC-Pipe: An Efficient Pipeline Scheme for Semi-honest MPC Machine Learning

Author(s): Y. Wang, R. Rajat, M. Annavaram

Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Feb 2025.

Paper Link

PIGEON: A High Throughput Framework for Private Inference of Neural Networks using Secure Multiparty Computation

Author(s): C. Harth-Kitzerow, Y. Wang, R. Rajat, G. Carle, M. Annavaram

The 25th Privacy Enhancing Technologies Symposium, Feb 2025.

Paper Link

Meta-Learn to Unlearn: Enhanced Exact Machine Unlearning in Recommendation Systems with Meta-Learning

Author(s): A. Alshabanah , K. Balasubramanian, M. Annavaram

The 25th Privacy Enhancing Technologies Symposium, Feb 2025.

Paper Link

Characterizing Context Influence and Hallucination in Summarization

Author(s): J. Flemings, W. Zhang, B. Jiang, Z. Takhirov, M. Annavaram

Neurips Workshop on Safe & Trustworthy Agents, Dec 2024.

Paper Link

Enabling Resource-Efficient On-Device Fine-Tuning of LLMs Using Only Inference Engines

Author(s): L. Gao, A. Ziashahabi, Y. Niu, S. Avestimehr, M. Annavaram

Neurips Workshop on Efficient Natural Language and Speech Processing, Dec 2024.

Paper Link

Biased User History Synthesis for Personalized Long-Tail Item Recommendation

Author(s): K. Balasubramanian, A. Alshabanah, E. Markowitz, G. Ver Steeg, M. Annavaram

Proceedings of the 18th ACM Conference on Recommender Systems, Oct 2024. (Acceptance rate 58/266 21.8%)

Paper Link

CADC: Encoding User-Item Interactions for Compressing Recommendation Model Training Data

Author(s): H. Entezari Zarch, A. Alshabanah, C. Jiang, M. Annavaram

Workshop on Risks, Opportunities, and Evaluation of Generative Models in Recommender Systems, Oct 2024.

Paper Link

High-Throughput Secure Multiparty Computation with an Honest Majority in Various Network Settings

Author(s): C. Harth-Kitzerow, A. Suresh, Y. Wang, H. Yalame, G. Carle, M. Annavaram

The 25th Privacy Enhancing Technologies Symposium, Aug 2024.

Paper Link

Differentially Private Knowledge Distillation via Synthetic Text Generation

Author(s): J. Flemings, M. Annavaram

The Findings of Association for Computational Linguistics, Aug 2024.

Paper Link

Differentially Private Next-Token Prediction of Large Language Models

Author(s): J. Flemings, M. Razaviyayn, M. Annavaram

North American Chapter of the Association for Computational Linguistics, Jun 2024. (Acceptance rate 565/2434, 23.2%)

Paper Link

Ethos: Rectifying Language Models in Orthogonal Parameter Space

Author(s): L. Gao, Y. Niu, T. Tang, S. Avestimehr, M. Annavaram

The Findings of North American Chapter of the Association for Computational Linguistics, Jun 2024. (Acceptance rate 304/2434, 12.5%)

Paper Link

Edge Private Graph Neural Networks with Singular Value Perturbation

Author(s): T. Tang, Y. Niu, S. Avestimehr, M. Annavaram

Proceedings of the 24th annual Privacy Enhancing Technologies Symposium (PETS), Jul 2024. (Acceptance rate 99/456, 22%)

Paper Link

LAORAM: Look Ahead ORAM for Recommendation Model Privacy

Author(s): Y. Wang, R. Rajat, M. Annavaram

Proceedings of the International Symposium on Computer Architecture (ISCA), Jun 2023. (Acceptance rate 79/373, 21%)

Paper Link

PageORAM: An Efficient DRAM Page Aware ORAM

Author(s): R. Rajat, Y. Wang, M. Annavaram

Proceedings of the 55th IEEE/ACM International Symposium on Microarchitecture, Oct 2022. (Acceptance rate 83/369, 22%)

Paper Link

StATIK: Structure and Text for Inductive Knowledge Graph Completion

Author(s): E.S. Markowitz, K. Balasubramanian, M. Mirtaheri, M. Annavaram, A. Galstyan, G.V. Steeg

2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Jul 2022.

Paper Link

Characterization of MPC-based Private Inferences for Transformer-based Models

Author(s): Y. Wang, E. Suh, W. Xiong, M. Annavaram, H. Lee

Proceedings of the International Symposium on Performance Analysis of Systems and Software, May 2022.

Paper Link

Enhancing Privacy Through Domain Adaptive Noise Injection for Speech Emotion Recognition

Author(s): T. Feng, H. Hashemi, M. Annavaram, S. Narayanan

Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, May 2022.

Paper Link

AVCC: Adaptative Verifiable Coded Computing

Author(s): T. Tang, H. Hashemi, R. Ali, S. Avestimehr, M. Annavaram

Proceedings of the International Conference on Parallel and Distributed Processing Systems, May 2022. (Acceptance rate 46/474, 10% Round 1 acceptance)

Paper Link

SpreadGNN: Serverless Multi-task Federated Learning for Molecular Graphs

Author(s): E. Ceyani, C. He, K. Balasubramaniyan, M. Annavaram, S. Avestimehr

Proceedings of the Thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI-22), Feb 2022. (Acceptance rate 1349/9251, 15%)

Paper Link

Check-N-Run: A Checkpointing System for Training Deep Learning Recommendation Models

Author(s): A. Eisenman, K. Matam, S. Ingram, D. Mudigere, R. Krishnamoorthi, K. Nair, M. Smelyanskiy, M. Annavaram

Proceedings of the 19th USENIX Symposium on Networked Systems Design and Implementation (NSDI), Apr 2022.

Paper Link

DarKnight: An Accelerated Framework for Privacy and Integrity Preserving Deep Learning Using Trusted Hardware

Author(s): H. Hashemi, Y. Wang, M. Annavaram

Proceedings of the 54th IEEE/ACM International Symposium on Microarchitecture, Oct 2021. (Acceptance rate 94/423, 22%)

Paper Link

cDLRM: Look Ahead Caching for Scalable Training of Recommendation Models

Author(s): K. Balasubramanian, A. Alsabnah, J. Choe, M. Annavaram

Proceedings of the 15th ACM Conference on Recommender Systems, Oct 2021. (Acceptance rate 49/267, 18%)

Paper Link

Origami Inference: Private Inference Using Hardware Enclaves

Author(s): K. Narra, Z. Lin, Y. Wang, K. Balasubramanian, M. Annavaram

IEEE International Conference on Cloud Computing (Short paper), Sep 2021. (Acceptance rate 23.8%)

Paper Link

MultiLogVC: Efficient Out-of-Core Graph Processing Framework on Flash Storage

Author(s): K. Matam, H. Hashemi, M. Annavaram

Proceedings of the International Conference on Parallel and Distributed Processing Systems, May 2021.

Paper Link

Group Knowledge Transfer: Federated Learning of Large CNNs at the Edge

Author(s): C. He, M. Annavaram, S. Avestimehr

Advances in Neural Information Processing Systems, Dec 2020. (Acceptance rate 1900/9454, 20%)

Paper Link

Collage Inference: Using Coded Redundancy for Lowering Latency Variation in Distributed Image Classification Systems

Author(s): K. Narra, Z. Lin, S. Avestimehr, G. Ananthanarayanan, M. Annavaram

Proceedings of the International Conference on Distributed Computing Systems, Jul 2020. (Acceptance rate 105/584, 18%)

Paper Link