Song Guo

Demystify Self-Attention in Vision Transformers from a Semantic Perspective: Analysis and Application

2022 · arXiv (Cornell University)

Self-attention mechanisms, especially multi-head self-attention (MSA), have achieved great success in many fields such as computer vision and natural language processing. However, many existing vision transformer (ViT) works simply inherent transformer designs from NLP to …

Non-Exemplar Online Class-incremental Continual Learning via Dual-prototype Self-augment and Refinement

2023 · arXiv (Cornell University)

This paper investigates a new, practical, but challenging problem named Non-exemplar Online Class-incremental continual Learning (NO-CL), which aims to preserve the discernibility of base classes without buffering data examples and efficiently learn novel classes continuously …

Online Federated Learning for Air-Ground Edge Intelligence

2023

Traditional federated learning (FL) usually relies on static ground base stations (BSs) for model aggregation. Unmanned aerial vehicles (UAVs), due to their flexible 3D deployment, can effectively complement ground BSs, enabling the establishment of air-ground …

BrokerChain: A Blockchain Sharding Protocol by Exploiting Broker Accounts

2024 · arXiv (Cornell University)

State-of-the-art blockchain sharding solutions such as Monoxide, can cause severely imbalanced distribution of transaction (TX) workloads across all blockchain shards due to the deployment policy of their accounts. Imbalanced TX distributions then produce hot shards, …

SecPQ: Secure Prediction Queries on Encrypted Outsourced Databases

2025 · IEEE Transactions on Dependable and Secure Computing

Prediction queries have revolutionized data search by integrating machine learning models and traditional data processing operations for advanced analytics. However, existing prediction query frameworks for outsourced databases face a critical security vulnerability: data flows are …

Mjölnir: Breaking the Shield of Perturbation-Protected Gradients via Adaptive Diffusion

2025 · Proceedings of the AAAI Conference on Artificial Intelligence

Perturbation-based mechanisms, such as differential privacy, mitigate gradient leakage attacks by introducing noise into the gradients, thereby preventing attackers from reconstructing clients' private data from the leaked gradients. However, can gradient perturbation protection mechanisms truly …

FreeMesh: Boosting Mesh Generation with Coordinates Merging

2025 · arXiv (Cornell University)

The next-coordinate prediction paradigm has emerged as the de facto standard in current auto-regressive mesh generation methods. Despite their effectiveness, there is no efficient measurement for the various tokenizers that serialize meshes into sequences. In …