Search

Your search keyword '"Zhu, Ziran"' showing total 127 results
Start Over Author "Zhu, Ziran"
127 results on '"Zhu, Ziran"'

2. Noise Dimension of GAN: An Image Compression Perspective

Author

Zhu, Ziran, Xu, Tongda, Li, Ling, and Wang, Yan

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Generative adversial network (GAN) is a type of generative model that maps a high-dimensional noise to samples in target distribution. However, the dimension of noise required in GAN is not well understood. Previous approaches view GAN as a mapping from a continuous distribution to another continous distribution. In this paper, we propose to view GAN as a discrete sampler instead. From this perspective, we build a connection between the minimum noise required and the bits to losslessly compress the images. Furthermore, to understand the behaviour of GAN when noise dimension is limited, we propose divergence-entropy trade-off. This trade-off depicts the best divergence we can achieve when noise is limited. And as rate distortion trade-off, it can be numerically solved when source distribution is known. Finally, we verifies our theory with experiments on image generation., Comment: ICME24

Published
2024

3. Consistency Model is an Effective Posterior Sample Approximation for Diffusion Inverse Solvers

Author

Xu, Tongda, Zhu, Ziran, Li, Jian, He, Dailan, Wang, Yuanyuan, Sun, Ming, Li, Ling, Qin, Hongwei, Wang, Yan, Liu, Jingjing, and Zhang, Ya-Qin

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

Diffusion Inverse Solvers (DIS) are designed to sample from the conditional distribution $p_{\theta}(X_0|y)$, with a predefined diffusion model $p_{\theta}(X_0)$, an operator $f(\cdot)$, and a measurement $y=f(x'_0)$ derived from an unknown image $x'_0$. Existing DIS estimate the conditional score function by evaluating $f(\cdot)$ with an approximated posterior sample drawn from $p_{\theta}(X_0|X_t)$. However, most prior approximations rely on the posterior means, which may not lie in the support of the image distribution, thereby potentially diverge from the appearance of genuine images. Such out-of-support samples may significantly degrade the performance of the operator $f(\cdot)$, particularly when it is a neural network. In this paper, we introduces a novel approach for posterior approximation that guarantees to generate valid samples within the support of the image distribution, and also enhances the compatibility with neural network-based operators $f(\cdot)$. We first demonstrate that the solution of the Probability Flow Ordinary Differential Equation (PF-ODE) with an initial value $x_t$ yields an effective posterior sample $p_{\theta}(X_0|X_t=x_t)$. Based on this observation, we adopt the Consistency Model (CM), which is distilled from PF-ODE, for posterior sampling. Furthermore, we design a novel family of DIS using only CM. Through extensive experiments, we show that our proposed method for posterior sample approximation substantially enhance the effectiveness of DIS for neural network operators $f(\cdot)$ (e.g., in semantic segmentation). Additionally, our experiments demonstrate the effectiveness of the new CM-based inversion techniques. The source code is provided in the supplementary material.

Published
2024

4. Variational Label-Correlation Enhancement for Congestion Prediction

Author

Liu, Biao, Qiao, Congyu, Xu, Ning, Geng, Xin, Zhu, Ziran, and Yang, Jun

Subjects
Computer Science - Machine Learning, Computer Science - Hardware Architecture
Abstract

The physical design process of large-scale designs is a time-consuming task, often requiring hours to days to complete, with routing being the most critical and complex step. As the the complexity of Integrated Circuits (ICs) increases, there is an increased demand for accurate routing quality prediction. Accurate congestion prediction aids in identifying design flaws early on, thereby accelerating circuit design and conserving resources. Despite the advancements in current congestion prediction methodologies, an essential aspect that has been largely overlooked is the spatial label-correlation between different grids in congestion prediction. The spatial label-correlation is a fundamental characteristic of circuit design, where the congestion status of a grid is not isolated but inherently influenced by the conditions of its neighboring grids. In order to fully exploit the inherent spatial label-correlation between neighboring grids, we propose a novel approach, {\ours}, i.e., VAriational Label-Correlation Enhancement for Congestion Prediction, which considers the local label-correlation in the congestion map, associating the estimated congestion value of each grid with a local label-correlation weight influenced by its surrounding grids. {\ours} leverages variational inference techniques to estimate this weight, thereby enhancing the regression model's performance by incorporating spatial dependencies. Experiment results validate the superior effectiveness of {\ours} on the public available \texttt{ISPD2011} and \texttt{DAC2012} benchmarks using the superblue circuit line.

Published
2023

10. Idempotence and Perceptual Image Compression

Author

Xu, Tongda, Zhu, Ziran, He, Dailan, Li, Yanghao, Guo, Lina, Wang, Yuanyuan, Wang, Zhe, Qin, Hongwei, Wang, Yan, Liu, Jingjing, Zhang, Ya-Qin, Xu, Tongda, Zhu, Ziran, He, Dailan, Li, Yanghao, Guo, Lina, Wang, Yuanyuan, Wang, Zhe, Qin, Hongwei, Wang, Yan, Liu, Jingjing, and Zhang, Ya-Qin

Abstract

Idempotence is the stability of image codec to re-compression. At the first glance, it is unrelated to perceptual image compression. However, we find that theoretically: 1) Conditional generative model-based perceptual codec satisfies idempotence; 2) Unconditional generative model with idempotence constraint is equivalent to conditional generative codec. Based on this newfound equivalence, we propose a new paradigm of perceptual image codec by inverting unconditional generative model with idempotence constraints. Our codec is theoretically equivalent to conditional generative codec, and it does not require training new models. Instead, it only requires a pre-trained mean-square-error codec and unconditional generative model. Empirically, we show that our proposed approach outperforms state-of-the-art methods such as HiFiC and ILLM, in terms of Fr\'echet Inception Distance (FID). The source code is provided in https://github.com/tongdaxu/Idempotence-and-Perceptual-Image-Compression., Comment: ICLR 2024

Published
2024

16. LAMPlace: Legalization-Aided Reinforcement Learning-Based Macro Placement for Mixed-Size Designs With Preplaced Blocks

Author

Gu, Hao, Gu, Jian, Peng, Keyu, Zhu, Ziran, Xu, Ning, Geng, Xin, and Yang, Jun

Abstract

Macro placement is a crucial and complex task in chip design. In recent years, reinforcement learning (RL) has emerged as a promising approach for autonomously generating effective macro placement on a grid-based chip canvas. However, existing RL-based macro placers encounter challenges when handling macros of varying sizes and preplaced blocks, due to using coarse-grained grids leads to wastage of placement space, while employing fine-grained grids results in an exponentially growing solution space and significant computational overhead. This brief proposes a legalization-aided RL-based macro placement (LAMPlace) for mixed-size designs with preplaced blocks. LAMPlace leverages RL to generate a global macro placement that allows possible overlap, followed by a macro legalization algorithm to ensure the legality of macro placement. LAMPlace can effectively address macros of various sizes and preplaced blocks because the action mask generation algorithm in RL is well coupled with the macro legalization algorithm. Compared with the state-of-the-art works, experimental results on publicly mixed-size placement benchmarks show that LAMPlace is effective.

Published
2024
Full Text
View/download PDF

24. An Effective Routing Refinement Algorithm Based on Incremental Replacement and Rerouting

Author

Zhu, Ziran, Shen, Fuheng, Mei, Yangjie, Chen, Jianli, and Yang, Jun

Abstract

To reduce the design complexity of physical design, placement and routing are commonly regarded as two problems. However, since the correlation and aims of placement and routing are not totally consistent, the complete separation of the two stages might limit the final routing quality. To narrow the gap between the two stages and improve the routing solution, this brief proposes an effective routing refinement algorithm based on incremental replacement and rerouting. We first propose an R-tree-based parallel net rerouting scheme to remove redundant routing segments, which offers more precise information for the subsequent cell replacement. Then, two location prediction methods are proposed and performed alternately in adjacent rounds of cell replacement. The first method predicts the cell locations sequentially based on the current routing results, while the second one adopts a gradient-descent-based method to predict the cell locations simultaneously from a more global perspective. Finally, we present an A*-based partial rerouting to generate the new routing results, followed by a simulated-annealing-based acceptance strategy to judge whether to accept the new routing results. Compared with the champion of the ICCAD 2020 contest and the state-of-the-art works, experimental results show that our algorithm is effective and efficient.

Published
2024
Full Text
View/download PDF

25. High-Performance 3-D Placement Engine With Physical-Aware Incremental Partitioning

Author

Zhu, Ziran, Shi, Yuejian, Mei, Yangjie, Shen, Fuheng, Liu, Hong, and Yang, Jun

Abstract

With the vertically stacked configuration and different technology processes imposed by stacked three-dimensional (3D) integration circuits (ICs), the placement problem has become more challenging. In this brief, we develop a high-performance 3D placement engine with physical-aware incremental partitioning for the stacked 3D ICs. We first propose a compact 2D global placement (GP) to find desired locations for all cells in a die, and then present a progressive density-driven partitioning to partition the netlist into two dies. After that, we propose an incremental half-perimeter wirelength (HPWL)-driven partitioning and a 3D GP to alternately optimize the partitioning and placement results. Finally, we present legalization and detailed placement methods to ensure the legality and further improve the solution. Compared with the top 3 teams of the ICCAD 2022 contest, experimental results on the contest benchmarks show that our algorithm achieves the best solution quality for every benchmark in a comparable runtime.

Published
2024
Full Text
View/download PDF

26. A Super Tough, Rapidly Biodegradable, Ultrafast Hemostatic Bioglue

Author

Wang, Hufei, primary, Cheng, Junyao, additional, Sun, Feifei, additional, Dou, Xueyu, additional, Liu, Jianheng, additional, Wang, Yiru, additional, Li, Ming, additional, Gao, Jianpeng, additional, Liu, Xiao, additional, Wang, Xing, additional, Yang, Fei, additional, Zhu, Ziran, additional, Shen, Hong, additional, Zhang, Licheng, additional, Tang, Peifu, additional, and Wu, Decheng, additional

Published
2023
Full Text
View/download PDF

49. Fast response and high sensitive photoelectrochemical ultraviolet detectors based on electrospinning SrTiO3 nanowires.

Author

Bai, Zhaowen, Zhu, Ziran, Wang, Gang, Wu, Hongchang, Wang, Youqing, Zhou, Jinyuan, Li, Jian, Pan, Xiaojun, and Xie, Erqing

Subjects
*ULTRAVIOLET detectors, *NANOWIRES, *ELECTROSPINNING, *SURFACE area, *CHARGE carriers
Abstract

One-dimensional nanowires can provide a direct electron transmission channel, as a result, the photogenerated carriers can be more easily and quickly transferred to the current collector. In a photoelectrochemical (PEC) ultraviolet (UV) detector, photoanode assemble using nanowires also possesses a large specific surface area to be beneficial for contacting between the electrolyte and photoanode. In this work, the SrTiO3 nanowires are fabricated by electrospinning and then assembled PEC UV detectors. Under UV illumination, the detector based on the SrTiO3 nanowires presents a superior detection performance with a response time of 8 ms. In particular, the device exhibits an efficient detection ability for weak UV light. Under the UV illumination intensity of 5 μW cm− 2, the current density Jsc reaches 11.3 nA cm− 2. The device also presents visible-blind, outstanding linear response and excellent repeatability under periodic UV irradiation. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results