7 papers
QATMA: Quantization-Aware Training with Multimodal Alignment for Open-Vocabulary Object Detection
Jinyeong Park, Donghwa Kang, Brent ByungHoon Kang +4
Quantizing open-vocabulary object detection (OVOD) models reduces their memory and computational costs, but extremely low-bit quantization severely degrades both cross-modal (regio…
SPOILER: TEE-Shielded DNN Partitioning of On-Device Secure Inference with Poison Learning
Donghwa Kang, Hojun Choe, Doohyun Kim +2
Deploying deep neural networks (DNNs) on edge devices exposes valuable intellectual property to model-stealing attacks. While TEE-shielded DNN partitioning (TSDP) mitigates this by…
Zero2Text: Zero-Training Cross-Domain Inversion Attacks on Textual Embeddings
Doohyun Kim, Donghwa Kang, Kyungjae Lee +2
The proliferation of retrieval-augmented generation (RAG) has established vector databases as critical infrastructure, yet they introduce severe privacy risks via embedding inversi…
STAS: Spatio-Temporal Adaptive Computation Time for Spiking Transformers
Donghwa Kang, Doohyun Kim, Sang-Ki Ko +3
Spiking neural networks (SNNs) offer energy efficiency over artificial neural networks (ANNs) but suffer from high latency and computational overhead due to their multi-timestep op…
Timestep-Compressed Attack on Spiking Neural Networks through Timestep-Level Backpropagation
Donghwa Kang, Doohyun Kim, Sang-Ki Ko +3
State-of-the-art (SOTA) gradient-based adversarial attacks on spiking neural networks (SNNs), which largely rely on extending FGSM and PGD frameworks, face a critical limitation: s…
CF-DETR: Coarse-to-Fine Transformer for Real-Time Object Detection
Woojin Shin, Donghwa Kang, Byeongyun Park +3
Detection Transformers (DETR) are increasingly adopted in autonomous vehicle (AV) perception systems due to their superior accuracy over convolutional networks. However, concurrent…