arXiv subjects

Hyojeong Lee

Publications and source records attributed to Hyojeong Lee.

6 recordsLinked to original sources

arXiv2026

GranQ: Efficient Channel-wise Quantization via Vectorized Pre-Scaling for Zero-Shot QAT

Zero-shot quantization (ZSQ) enables neural network compression without original training data, making it a promising solution for restricted data access scenarios. To compensate for the lack of data, recent ZSQ methods typically rely on synthetic inputs generated from the full-precision model. However, these synthetic inputs often lead to activation distortion, especially under low-bit settings. To mitigate this, existing methods typically employ per-channel scaling, but they still struggle due to the severe computational overhead during the accumulation process. To overcome this critical bottleneck, we propose GranQ, a novel activation quantization framework that introduces an efficient pre-scaling strategy. Unlike conventional channel-wise methods that repeatedly perform scaling operations during accumulation, GranQ applies scaling factors in a pre-scaling step through fully vectorized computation, eliminating runtime scaling overhead. This design enables GranQ to maintain fine-grained quantization accuracy while significantly reducing computational burden, particularly in low-bit quantization settings. Extensive experiments under quantization-aware training (QAT) settings demonstrate that GranQ consistently outperforms state-of-the-art ZSQ methods across CIFAR and ImageNet. In particular, our method achieves up to 5.45% higher accuracy in the 3-bit setting on CIFAR-100 and even surpasses the full-precision baseline on CIFAR-10.

cs.CV↗

arXiv2026

Rethinking Token Reduction for Diffusion Models via Output-Similarity-Awareness

Diffusion Transformers (DiTs) achieve superior image generation quality but suffer from quadratic computational complexity relative to token count. While various token reduction (TR) methods have been proposed to mitigate this cost, they overlook the primary objective of generative models: minimizing recovery error, which requires reflecting output token similarity. They rely solely on input token similarity inherited from reduction-only ViT paradigms, leading to a fundamental misalignment with this objective. To bridge this gap, we propose DiTo, a novel TR paradigm that shifts the focus toward output-centric token reduction. Based on the observation that output token similarity is consistently preserved across adjacent timesteps, DiTo utilizes prior-step similarities as an effective proxy to establish token correspondences at a Matching timestep, which are then reused across multiple subsequent Reduction timesteps. To optimize this interleaved scheduling, we propose Pair Match Ratio (PMR)-guided Interval Scheduling to determine the optimal matching frequency. Furthermore, to mitigate localized approximation errors and resulting blocking artifacts caused by repeated reuse, we propose Frequency-aware Token Matching by incorporating a selection-frequency penalty. Extensive experiments demonstrate that DiTo consistently outperforms existing TR methods with 1.6-3.9 dB higher PSNR at comparable speedups, achieving a superior Pareto frontier.

cs.CV↗

arXiv2025

ABC-FHE : A Resource-Efficient Accelerator Enabling Bootstrappable Parameters for Client-Side Fully Homomorphic Encryption

As the demand for privacy-preserving computation continues to grow, fully homomorphic encryption (FHE)-which enables continuous computation on encrypted data-has become a critical solution. However, its adoption is hindered by significant computational overhead, requiring 10000-fold more computation compared to plaintext processing. Recent advancements in FHE accelerators have successfully improved server-side performance, but client-side computations remain a bottleneck, particularly under bootstrappable parameter configurations, which involve combinations of encoding, encrypt, decoding, and decrypt for large-sized parameters. To address this challenge, we propose ABC-FHE, an area- and power-efficient FHE accelerator that supports bootstrappable parameters on the client side. ABC-FHE employs a streaming architecture to maximize performance density, minimize area usage, and reduce off-chip memory access. Key innovations include a reconfigurable Fourier engine capable of switching between NTT and FFT modes. Additionally, an on-chip pseudo-random number generator and a unified on-the-fly twiddle factor generator significantly reduce memory demands, while optimized task scheduling enhances the CKKS client-side processing, achieving reduced latency. Overall, ABC-FHE occupies a die area of 28.638 mm2 and consumes 5.654 W of power in 28 nm technology. It delivers significant performance improvements, achieving a 1112x speed-up in encoding and encryption execution time compared to a CPU, and 214x over the state-of-the-art client-side accelerator. For decoding and decryption, it achieves a 963x speed-up over the CPU and 82x over the state-of-the-art accelerator.

cs.AR↗

arXiv2025

Advanced Knowledge Transfer: Refined Feature Distillation for Zero-Shot Quantization in Edge Computing

We introduce AKT (Advanced Knowledge Transfer), a novel method to enhance the training ability of low-bit quantized (Q) models in the field of zero-shot quantization (ZSQ). Existing research in ZSQ has focused on generating high-quality data from full-precision (FP) models. However, these approaches struggle with reduced learning ability in low-bit quantization due to its limited information capacity. To overcome this limitation, we propose effective training strategy compared to data generation. Particularly, we analyzed that refining feature maps in the feature distillation process is an effective way to transfer knowledge to the Q model. Based on this analysis, AKT efficiently transfer core information from the FP model to the Q model. AKT is the first approach to utilize both spatial and channel attention information in feature distillation in ZSQ. Our method addresses the fundamental gradient exploding problem in low-bit Q models. Experiments on CIFAR-10 and CIFAR-100 datasets demonstrated the effectiveness of the AKT. Our method led to significant performance enhancement in existing generative models. Notably, AKT achieved significant accuracy improvements in low-bit Q models, achieving state-of-the-art in the 3,5bit scenarios on CIFAR-10. The code is available at https://github.com/Inpyo-Hong/AKT-Advanced-knowledge-Transfer.

cs.CV↗

arXiv2024

MRNet: Multifaceted Resilient Networks for Medical Image-to-Image Translation

We propose a Multifaceted Resilient Network(MRNet), a novel architecture developed for medical image-to-image translation that outperforms state-of-the-art methods in MRI-to-CT and MRI-to-MRI conversion. MRNet leverages the Segment Anything Model (SAM) to exploit frequency-based features to build a powerful method for advanced medical image transformation. The architecture extracts comprehensive multiscale features from diverse datasets using a powerful SAM image encoder and performs resolution-aware feature fusion that consistently integrates U-Net encoder outputs with SAM-derived features. This fusion optimizes the traditional U-Net skip connection while leveraging transformer-based contextual analysis. The translation is complemented by an innovative dual-mask configuration incorporating dynamic attention patterns and a specialized loss function designed to address regional mapping mismatches, preserving both the gross anatomy and tissue details. Extensive validation studies have shown that MRNet outperforms state-of-the-art architectures, particularly in maintaining anatomical fidelity and minimizing translation artifacts.

eess.IV↗

arXiv2022

Laser pulse compression by a density gradient plasma for exawatt to zettawatt lasers

We propose a new method for laser pulse compression that uses the spatially varying dispersion of a plasma plume with a density gradient. This novel scheme can be used to compress ultrahigh power lasers. A long, negatively frequency-chirped, laser pulse reflects off the plasma ramp of an over-dense plasma. As the density increases longitudinally the high frequency photons at the leading part of the laser pulse propagates more deeply than low frequency photons, the pulse is compressed in a similar way to compression off a chirped mirror. Proof-of-principle simulations, using a one-dimensional (1-D) particle-in-cell (PIC) simulation code demonstrates the compression of 2.35 ps laser pulse to 10.3 fs, with a compression ratio of 225. As plasmas is robust and resistant to high intensities, unlike gratings in a chirped-pulse amplification (CPA) technique [1], the method could be used as a compressor to reach exawatt or zettawatt peak power lasers.

physics.plasm-ph↗