Filters








12,427 Hits in 3.2 sec

Dynamical properties of the Haldane chain with bond disorder [article]

Jing-Kai Fang, Jun-Han Huang, Han-Qing Wu, Dao-Xin Yao
2021 arXiv   pre-print
ACKNOWLEDGMENTSFIG. 8 . 8 We thank Yu-Cheng Lin, Anders W.Sandvik, and Muwei Wu for helpful discussions.  ... 
arXiv:2109.00552v1 fatcat:ojl2sn7xgrfefnakqbixilyagu

Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection [article]

Jianyuan Guo, Kai Han, Yunhe Wang, Chao Zhang, Zhaohui Yang, Han Wu, Xinghao Chen, Chang Xu
2020 arXiv   pre-print
Neural Architecture Search (NAS) has achieved great success in image classification task. Some recent works have managed to explore the automatic design of efficient backbone or feature fusion layer for object detection. However, these methods focus on searching only one certain component of object detector while leaving others manually designed. We identify the inconsistency between searched component and manually designed ones would withhold the detector of stronger performance. To this end,
more » ... e propose a hierarchical trinity search framework to simultaneously discover efficient architectures for all components (i.e. backbone, neck, and head) of object detector in an end-to-end manner. In addition, we empirically reveal that different parts of the detector prefer different operators. Motivated by this, we employ a novel scheme to automatically screen different sub search spaces for different components so as to perform the end-to-end search for each component on the corresponding sub search space efficiently. Without bells and whistles, our searched architecture, namely Hit-Detector, achieves 41.4\% mAP on COCO minival set with 27M parameters. Our implementation is available at https://github.com/ggjy/HitDet.pytorch.
arXiv:2003.11818v1 fatcat:udwiiu4s2nchtblhxjqdauflk4

Generation of ice states through deep reinforcement learning [article]

Kai-Wen Zhao, Wen-Han Kao, Kai-Hsin Wu, Ying-Jer Kao
2019 arXiv   pre-print
We present a deep reinforcement learning framework where a machine agent is trained to search for a policy to generate a ground state for the square ice model by exploring the physical environment. After training, the agent is capable of proposing a sequence of local moves to achieve the goal. Analysis of the trained policy and the state value function indicates that the ice rule and loop-closing condition are learned without prior knowledge. We test the trained policy as a sampler in the
more » ... chain Monte Carlo and benchmark against the baseline loop algorithm. This framework can be generalized to other models with topological constraints where generation of constraint-preserving states is difficult.
arXiv:1903.04698v1 fatcat:2lm6gbl2sjb7xpzdjhkasb65ry

CMT: Convolutional Neural Networks Meet Vision Transformers [article]

Jianyuan Guo, Kai Han, Han Wu, Chang Xu, Yehui Tang, Chunjing Xu, Yunhe Wang
2021 arXiv   pre-print
Vision transformers have been successfully applied to image recognition tasks due to their ability to capture long-range dependencies within an image. However, there are still gaps in both performance and computational cost between transformers and existing convolutional neural networks (CNNs). In this paper, we aim to address this issue and develop a network that can outperform not only the canonical transformers, but also the high-performance convolutional models. We propose a new transformer
more » ... based hybrid network by taking advantage of transformers to capture long-range dependencies, and of CNNs to model local features. Furthermore, we scale it to obtain a family of models, called CMTs, obtaining much better accuracy and efficiency than previous convolution and transformer based models. In particular, our CMT-S achieves 83.5% top-1 accuracy on ImageNet, while being 14x and 2x smaller on FLOPs than the existing DeiT and EfficientNet, respectively. The proposed CMT-S also generalizes well on CIFAR10 (99.2%), CIFAR100 (91.7%), Flowers (98.7%), and other challenging vision datasets such as COCO (44.3% mAP), with considerably less computational cost.
arXiv:2107.06263v2 fatcat:clxhk6spefhjzdqyf4kkn4q23a

Expanding Sparse Guidance for Stereo Matching [article]

Yu-Kai Huang, Yueh-Cheng Liu, Tsung-Han Wu, Hung-Ting Su, Winston H. Hsu
2020 arXiv   pre-print
The performance of image based stereo estimation suffers from lighting variations, repetitive patterns and homogeneous appearance. Moreover, to achieve good performance, stereo supervision requires sufficient densely-labeled data, which are hard to obtain. In this work, we leverage small amount of data with very sparse but accurate disparity cues from LiDAR to bridge the gap. We propose a novel sparsity expansion technique to expand the sparse cues concerning RGB images for local feature
more » ... ment. The feature enhancement method can be easily applied to any stereo estimation algorithms with cost volume at the test stage. Extensive experiments on stereo datasets demonstrate the effectiveness and robustness across different backbones on domain adaption and self-supervision scenario. Our sparsity expansion method outperforms previous methods in terms of disparity by more than 2 pixel error on KITTI Stereo 2012 and 3 pixel error on KITTI Stereo 2015. Our approach significantly boosts the existing state-of-the-art stereo algorithms with extremely sparse cues.
arXiv:2005.02123v1 fatcat:fqyjjkumczhdlazzvk7mt6uha4

GhostNets on Heterogeneous Devices via Cheap Operations [article]

Kai Han, Yunhe Wang, Chang Xu, Jianyuan Guo, Chunjing Xu, Enhua Wu, Qi Tian
2022 arXiv   pre-print
Han et.al. (Han et al., 2016) proposed to prune the unimportant weights in neural networks. (Li et al., 2017) utilized LOGIES CO., LTD.  ...  Pruning connections (Han et al., 2015 (Han et al., , 2016;; Wang et al., 2016) cuts out the unimportant connections between neurons.  ... 
arXiv:2201.03297v1 fatcat:ckmt7wh3vvfiphco5goyteyyji

Transformer in Transformer [article]

Kai Han, An Xiao, Enhua Wu, Jianyuan Guo, Chunjing Xu, Yunhe Wang
2021 arXiv   pre-print
Transformer is a new kind of neural architecture which encodes the input data as powerful features via the attention mechanism. Basically, the visual transformers first divide the input images into several local patches and then calculate both representations and their relationship. Since natural images are of high complexity with abundant detail and color information, the granularity of the patch dividing is not fine enough for excavating features of objects in different scales and locations.
more » ... n this paper, we point out that the attention inside these local patches are also essential for building visual transformers with high performance and we explore a new architecture, namely, Transformer iN Transformer (TNT). Specifically, we regard the local patches (e.g., 16×16) as "visual sentences" and present to further divide them into smaller patches (e.g., 4×4) as "visual words". The attention of each word will be calculated with other words in the given visual sentence with negligible computational costs. Features of both words and sentences will be aggregated to enhance the representation ability. Experiments on several benchmarks demonstrate the effectiveness of the proposed TNT architecture, e.g., we achieve an 81.5 higher than that of the state-of-the-art visual transformer with similar computational cost. The PyTorch code is available at https://github.com/huawei-noah/CV-Backbones, and the MindSpore code is available at https://gitee.com/mindspore/models/tree/master/research/cv/TNT.
arXiv:2103.00112v3 fatcat:i42cw3eebraa5b4cm6axpkazge

Hire-MLP: Vision MLP via Hierarchical Rearrangement [article]

Jianyuan Guo, Yehui Tang, Kai Han, Xinghao Chen, Han Wu, Chao Xu, Chang Xu, Yunhe Wang
2021 arXiv   pre-print
Previous vision MLPs such as MLP-Mixer and ResMLP accept linearly flattened image patches as input, making them inflexible for different input sizes and hard to capture spatial information. Such approach withholds MLPs from getting comparable performance with their transformer-based counterparts and prevents them from becoming a general backbone for computer vision. This paper presents Hire-MLP, a simple yet competitive vision MLP architecture via Hierarchical rearrangement, which contains two
more » ... evels of rearrangements. Specifically, the inner-region rearrangement is proposed to capture local information inside a spatial region, and the cross-region rearrangement is proposed to enable information communication between different regions and capture global context by circularly shifting all tokens along spatial directions. Extensive experiments demonstrate the effectiveness of Hire-MLP as a versatile backbone for various vision tasks. In particular, Hire-MLP achieves competitive results on image classification, object detection and semantic segmentation tasks, e.g., 83.8 box AP and 44.8 previous transformer-based and MLP-based models with better trade-off for accuracy and throughput. Code is available at https://github.com/ggjy/Hire-Wave-MLP.pytorch.
arXiv:2108.13341v2 fatcat:shwg5l7lszebxlkcoaxluco74e

Robotic-assisted right middle lobectomy

Xingshi Chen, Su Yang, Wei Guo, Runsen Jin, Yajie Zhang, Han Wu, Hailei Du, Dingpei Han, Kai Chen, Jie Xiang, Hecheng Li
2017 AME Medical Journal  
A ground-glass opacity (GGO) was detected in the medial segment of the right middle lung lobe of a 48-year-old man about one year ago during a regular medical examination. He did not have any symptoms (e.g., chest tightness, shortness of breath, cough, or expectoration) at that time; therefore, we suggested he undergo regular re-examinations. Result of a computedtomography (CT) scan one month ago still indicated a 0.9-cm GGO in the same position. The patient still had no symptoms. Positron
more » ... ion tomography-CT showed hypermetabolism in this nodule. The patient's physical performance was good, and his appetite, sleep, urination, and defecation were normal. His body weight did not change significantly. He denied smoking or alcohol abuse, and family history. Physical examination A complete physical examination was performed when the patient was admitted. No abnormity was found. No lymph nodes were palpable in the neck, axilla, or below the clavicle. Auxiliary examination Chest CT showed a 1.1-cm elliptic GGO in the right middle lung lobe, with clear margins, without lobulation sign, spiculation or vacuole sign (Figure 1) . No abnormity was found in the lung hila. No abnormal lymph nodes were found in the mediastinum. PET/CT showed hypermetabolism in the nodule of the right middle lung lobe. Abdominal ultrasound scan, bone scan, cranial magnetic resonance imaging, echocardiogram, and pulmonary Case Report Abstract: Da Vinci surgical system has been widely used in thoracic surgery as a new technique. We are going to share the experience of robotic surgery for right middle lobectomy. A 48-year-old patient with a ground-glass opacity (GGO) underwent robotic-assisted right middle lobectomy in our center. The patient was discharged on postoperative day 4 without any perioperative complications. The pathological stage was T1aN0M0 (stage IA). -assisted right middle lobectomy. AME Med J 2017;2:8. from the patient for publication of this manuscript and any accompanying images.
doi:10.21037/amj.2017.01.07 fatcat:6yzg6rpeb5dqvlvwurfnu53jgq

Dynamic Resolution Network [article]

Mingjian Zhu, Kai Han, Enhua Wu, Qiulin Zhang, Ying Nie, Zhenzhong Lan, Yunhe Wang
2021 arXiv   pre-print
Wu et.al. [35] introduces BlockDrop which learns to dynamically execute the necessary layers so as to best reduce total computation.  ...  Revealing pattern redundancy among feature maps, Han [5] proposes to generate more feature maps from intrinsic ones through some cheap operations and Zhang et.al. [41] adopts relatively heavy computation  ... 
arXiv:2106.02898v3 fatcat:njmt2yyaifgcnat72yhismjlvi

Oxidative alkoxycarbonylation of terminal alkenes with carbazates

Yu-Han Su, Zhao Wu, Shi-Kai Tian
2013 Chemical Communications  
Wu and S.-K. Tian, Eur. J. Org. Chem., 2012, 1590. 3 P. Wang, C.-R. Liu, X.-L. Sun, S.-S. Chen, J.-F. Li, Z. Xie and Y. Tang, Chem. Commun., 2012, 48, 290. 4 C. Ramesh, G. Mahender, N.  ... 
doi:10.1039/c3cc42994f pmid:23760054 fatcat:ao6su4hllvfoljonmtizfhyr34

Training Binary Neural Networks through Learning with Noisy Supervision [article]

Kai Han, Yunhe Wang, Yixing Xu, Chunjing Xu, Enhua Wu, Chang Xu
2020 arXiv   pre-print
Correspondence to: Yunhe Wang <yunhe.wang@huawei.com>, Enhua Wu <ehwu@umac.mo>, Chang Xu <c.xu@sydney.edu.au>.  ...  A series of model compression and acceleration methods (Han et al., 2016; Chen et al., 2020) have been proposed to reduce the number of parameters and FLOPs of CNNs, including network pruning (Han et  ... 
arXiv:2010.04871v1 fatcat:oparhf54gnh3ljldb2p3dj3z24

Homocysteine level, body mass index and clinical correlates in Chinese Han patients with schizophrenia

Yuanyuan Huang, Kai Wu, Hehua Li, Jing Zhou, Dongsheng Xiong, Xia Huang, Jiahui Li, Ya Liu, Zhilin Pan, David T. Mitchell, Fengchun Wu, Xiang Yang Zhang
2020 Scientific Reports  
This cross-sectional naturalistic study aimed to evaluate the relationship between BMI, Hcy and clinical symptoms in Chinese Han patients with chronic schizophrenia.  ...  data, glucose and lipid metabolism parameters between high and low BMI groups in patients with schizophrenia; (2) whether there were a correlation between BMI, Hcy and clinical symptoms in Chinese Han  ...  Inclusion criteria included the following: (1) aged 18-60 years, Han Chinese; (2) confirmed DSM-V diagnostic criteria of schizophrenia by two experienced psychiatrists according to the Structured Clinical  ... 
doi:10.1038/s41598-020-72934-3 pmid:32999343 fatcat:vtqwiwtcffdttjkfqijazfmpu4

Deterministic Approximation for Submodular Maximization over a Matroid in Nearly Linear Time [article]

Kai Han, Zongmai Cao, Shuang Cui, Benwei Wu
2020 arXiv   pre-print
We study the problem of maximizing a non-monotone, non-negative submodular function subject to a matroid constraint. The prior best-known deterministic approximation ratio for this problem is 1/4-ϵ under 𝒪((n^4/ϵ)log n) time complexity. We show that this deterministic ratio can be improved to 1/4 under 𝒪(nr) time complexity, and then present a more practical algorithm dubbed TwinGreedyFast which achieves 1/4-ϵ deterministic ratio in nearly-linear running time of 𝒪(n/ϵlogr/ϵ). Our approach is
more » ... ed on a novel algorithmic framework of simultaneously constructing two candidate solution sets through greedy search, which enables us to get improved performance bounds by fully exploiting the properties of independence systems. As a byproduct of this framework, we also show that TwinGreedyFast achieves 1/2p+2-ϵ deterministic ratio under a p-set system constraint with the same time complexity. To showcase the practicality of our approach, we empirically evaluated the performance of TwinGreedyFast on two network applications, and observed that it outperforms the state-of-the-art deterministic and randomized algorithms with efficient implementations for our problem.
arXiv:2010.11420v1 fatcat:likipanpybcfra56r5zcgp6dfy

Pierce and push : A simplified method to facilitate laparoscopic myomectomy

Chin-Jung Wang, Chyi-Long Lee, Pei-Ju Wu, Kai-Yun Wu, Chien-Min Han, Hsuan Su
2012 Biomedical Journal  
The aim of this study was to present our experience with a novel approach called the pierce and push (PP) method which uses a stainless steel centimeter probe instead of a 5 mm claw forceps to enucleate fibroids in a laparoscopic myomectomy (LM). Methods: A retrospective chart review of 90 women with symptomatic uterine fibroids who underwent an LM was performed. Cases of LM with the PP method were compared with a matched control group of LM with a 5 mm claw forceps. The operative time, tumor
more » ... paration time, specimen removal time, amount of blood loss, requirement for blood transfusion and length of hospital stay were compared between groups. Results: The two groups were matched by age, body mass index, previous cesarean delivery, main fibroid size, and number and weight of fibroids. The tumor separation time was significantly shorter in the PP group than the claw forceps group (9.7 3.1 minutes versus 17.1 4.4 minutes, p < 0.001). The length of the operation, hospitalization time, specimen removal time, amount of blood loss, and requirement for blood transfusion were not significantly different between groups. Conclusion: A stainless steel centimeter probe has an advantage over a 5 mm claw forceps in pushing and pulling fibroids. Our findings indicate that the PP method was much more effective in excision of fibroids than a 5 mm claw grasper in LM. (Chang Gung Med J 2012;35:160-6)
doi:10.4103/2319-4170.106155 fatcat:3omz2lrf2jgkbizldvun7lvtoa
« Previous Showing results 1 — 15 out of 12,427 results