检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Guan, Jian Huang, Minlie The CoAI Group DCST Institute for Artificial Intelligence State Key Lab of Intelligent Technology and Systems Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China

Despite the huge progress in a myriad of generation tasks, pretrained language models (LMs) such as GPT2 still tend to generate repetitive texts with maximization-based decoding algorithms for open-ended generation. We attribute their overestimation of token-level repetition probabilities to the learning bias: LMs capture simple repetitive patterns faster with the MLE loss. We propose self-contrastive training to penalize the output of a premature checkpoint of the same model when it incorrectly predicts repetition, which is shown to mitigate repetition effectively while maintaining fluency on two datasets. Furthermore, we find that LMs use longer-range dependencies to predict repetitive tokens than non-repetitive ones, which may be the cause of sentence-level repetition loops. Copyright © 2023, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ETHICIST: Targeted Training Data Extraction Through Loss Smoothed Soft Prompting and Calibrated Confidence Estimation

arXiv

引用

arXiv 2023年

作者： Zhang, Zhexin Wen, Jiaxin Huang, Minlie The CoAI group DCST Institute for Artificial Intelligence State Key Lab of Intelligent Technology and Systems Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China

Large pre-trained language models achieve impressive results across many tasks. However, recent works point out that pre-trained language models may memorize a considerable fraction of their training data, leading to the privacy risk of information leakage. In this paper, we propose a method named ETHICIST for targeted training data Extraction THrough loss smoothed soft prompting and calIbrated ConfIdence eSTimation, investigating how to recover the suffix in the training data when given a prefix. To elicit memorization in the attacked model, we tune soft prompt embeddings while keeping the model fixed. We further propose a smoothing loss that smooths the loss distribution of the suffix tokens to make it easier to sample the correct suffix. In order to select the most probable suffix from a collection of sampled suffixes and estimate the prediction confidence, we propose a calibrated confidence estimation method, which normalizes the confidence of the generated suffixes with a local estimation. We show that ETHICIST significantly improves the extraction performance on a recently proposed public benchmark. We also investigate several factors influencing the data extraction performance, including decoding strategy, model scale, prefix length, and suffix length. Our code is availab.e at https://***/thu-coai/Targeted-Data-Extraction. Copyright © 2023, The Authors. All rights reserved.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

Optimized 25Gb/s Electroabsorption Modulated Tunable Lasers for O-band DWDM systems 20

Optimized 25Gb/s Electroabsorption Modulated Tunable Lasers ...

引用

20th International Conference on Optical Communications and Networks, ICOCN 2022

作者： Liu, Hao Sun, Zhenxing Wang, Lei Xiao, Rulei Zhang, Anxu Lv, Kai Feng, Lipeng Deng, Ji Huo, Xiaoli China Telecom Research Institute Beijing China College of Engineering and Applied Sciences and Institute of Optical Communication Engineering Nanjing University Key Lab. of Intelligent Optical Sensing and Manipulation of the Min. of Educ. and Natl. Lab. of Solid-State Microstructures Nanjing China Wuxi Taclink Optoelectronics Technology Co Ltd Wuxi China

ISBN: (数字)9781665458986

ISBN: (纸本)9781665458986

We propose and experimentally demonstrate an O-band electroabsorption modulated tunable in-series DFB laser array (EMTL). Four lasers with 400-GHz spacing in the laser array are placed in series to avoid the usage of an optical combiner and the additional power loss. The reconstruction-equivalent-chirp technique is used to simplify the grating fabrication and precisely control the grating phase. A semiconductor optical amplifier is integrated in front of the electroabsorption modulator (EAM) to amplify the output power. As a result, we realized 16 channels with 100 GHz spacing with the SMSRs of above 40 dB. The modulation bandwidth of the EAM is above 20 GHz, and 25Gb/s data per wavelength is obtained. The proposed EMTL shows the potential application as a compact and low-cost light source for DWDM systems, such as the next-generation fronthaul networks and passive optical networks in O-band. © 2022 IEEE.

关键词： Laser tuning

来源：评论

学校读者我要写书评

暂无评论

DA-Ada: learning domain-aware adapter for domain adaptive object detection 24

DA-Ada: learning domain-aware adapter for domain adaptive ob...

引用

Proceedings of the 38th International Conference on Neural Information Processing systems

作者： Haochen Li Rui Zhang Hantao Yao Xin Zhang Yifan Hao Xinkai Song Xiaqing Li Yongwei Zhao Ling Li Yunji Chen Intelligent Software Research Center Institute of Software CAS Beijing China and University of Chinese Academy of Sciences Beijing China State Key Lab of Processors Institute of Computing Technology CAS Beijing China State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation CAS Beijing China State Key Lab of Processors Institute of Computing Technology CAS Beijing China and University of Chinese Academy of Sciences Beijing China

ISBN: (纸本)9798331314385

Domain adaptive object detection (DAOD) aims to generalize detectors trained on an annotated source domain to an unlab.lled target domain. As the visual-language models (VLMs) can provide essential general knowledge on unseen images, freezing the visual encoder and inserting a domain-agnostic adapter can learn domaininvariant knowledge for DAOD. However, the domain-agnostic adapter is inevitably biased to the source domain. It discards some beneficial knowledge discriminative on the unlab.lled domain, i.e. domain-specific knowledge of the target domain. To solve the issue, we propose a novel Domain-Aware Adapter (DA-Ada) tailored for the DAOD task. The key point is exploiting domain-specific knowledge between the essential general knowledge and domain-invariant knowledge. DA-Ada consists of the Domain-Invariant Adapter (DIA) for learning domain-invariant knowledge and the Domain-Specific Adapter (DSA) for injecting the domain-specific knowledge from the information discarded by the visual encoder. Comprehensive experiments over multiple DAOD tasks show that DA-Ada can efficiently infer a domain-aware visual encoder for boosting domain adaptive object detection. Our code is availab.e at https://***/Therock90421/DA-Ada.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prototype-based classifier learning for long-tailed visual recognition

引用

Science China(Information Sciences) 2022年第6期65卷 62-76页

作者： Xiu-Shen WEI Shu-Lin XU Hao CHEN Liang XIAO Yuxin PENG School of Computer Science and Engineering Nanjing University of Science and Technology State Key Laboratory of Integrated Services Networks Xidian University Key Lab of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education and Jiangsu Key Lab of Image and Video Understanding for Social Security Wangxuan Institute of Computer Technology Peking University

In this paper, we tackle the long-tailed visual recognition problem from the categorical prototype perspective by proposing a prototype-based classifier learning(PCL) method. Specifically, thanks to the generalization ability and robustness, categorical prototypes reveal their advantages of representing the category semantics. Coupled with their class-balance characteristic, categorical prototypes also show potential for handling data imbalance. In our PCL, we propose to generate the categorical classifiers based on the prototypes by performing a learnable mapping function. To further alleviate the impact of imbalance on classifier generation, two kinds of classifier calibration approaches are designed from both prototype-level and example-level aspects. Extensive experiments on five benchmark datasets, including the large-scale iNaturalist, Places-LT,and ImageNet-LT, justify that the proposed PCL can outperform state-of-the-arts. Furthermore, validation experiments can demonstrate the effectiveness of tailored designs in PCL for long-tailed problems.

关键词： long-tailed distribution categorical prototype classifier generation classifier calibration class imbalance

来源：评论

学校读者我要写书评

暂无评论

Coordinated Navigation Control of Cross-Domain Unmanned systems via Guiding Vector Fields

arXiv

引用

arXiv 2023年

作者： Hu, Bin-Bin Zhang, Hai-Tao Liu, Bin Ding, Jianing Xu, Yifan Luo, Chuanshang Cao, Haosen The School of Artificial Intelligence and Automation The Engineering Research Center of Autonomous Intelligent Unmanned Systems The Key Laboratory of Image Processing and Intelligent Control The State Key Lab of Digital Manufacturing Equipment and Technology Huazhong University of Science and Technology Wuhan430074 China Dongguan University of Technology Dongguan523106 China

This paper proposes a distributed guiding-vector-field (DGVF) controller for cross-domain unmanned systems (CDUSs) consisting of heterogeneous unmanned aerial vehicles (UAVs) and unmanned surface vehicles (USVs), to achieve coordinated navigation whereas maneuvering along their prescribed paths. In particular, the DGVF controller provides a hierarchical architecture of an upper-level heterogeneous guidance velocity controller and a lower-level signal tracking regulator. Therein, the upper-level controller is to govern multiple heterogeneous USVs and UAVs to approach and maneuver along the prescribed paths and coordinate the formation simultaneously, whereas the low-level regulator is to track the corresponding desired guidance signals provided by the upper-level module. Significantly, the heterogeneous coordination among neighboring UAVs and USVs is achieved merely by the lightweight communication of a scalar (i.e., the additional virtual coordinate), which substantially decreases the communication and computational costs. Sufficient conditions assuring asymptotical convergence of the closed-loop system are derived in presence of the exponentially vanishing tracking errors. Finally, real-lake experiments are conducted on a self-established cross-domain heterogeneous platform consisting of three M-100 UAVs, two HUSTER-16 USVs, a HUSTER-12C USV, and a WiFi 5G wireless communication station to verify the effectiveness of the present DGVF controller. Copyright © 2023, The Authors. All rights reserved.

关键词： Air navigation

来源：评论

学校读者我要写书评

暂无评论

Design of Simultaneous state and Unknown Input Filtering Algorithm for a Class of Nonlinear Stochastic systems with Multiple Sensors 8

Design of Simultaneous State and Unknown Input Filtering Alg...

引用

8th IEEE International Conference on Cloud Computing and Intelligence systems, CCIS 2022

作者： Hu, Zhibin Hu, Jun Chen, Cai Du, Junhua Harbin University of Science and Technology Department of Applied Mathematics Harbin150080 China Harbin University of Science and Technology School of Automation Harbin150080 China Harbin University of Science and Technology Heilongjiang Prov. Key Lab. of Optimization Control and Intelligent Analysis for Complex Systems Harbin150080 China Qiqihar University College of Science Qiqihar161006 China

ISBN: (纸本)9781665477352

In this paper, we investigate the simultaneous state and unknown input (SUI) filtering issue for a class of multi-sensor networked systems (MSNSs). The unknown input with no prior knowledge is introduced in the system state and output. The focus of this paper is to design the local filters with regard to the SUI, which can yield that the local upper bounds of the filtering error covariance for the SUI are derived at each instant. Moreover, the local filter gains of the SUI are designed such that the obtained upper bounds can be minimized. Finally, the proposed joint SUI algorithm is verified by using the simulation example. © 2022 IEEE.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

Marker or Markerless? Mode-Switchable Optical Tactile Sensing for Diverse Robot Tasks

arXiv

引用

arXiv 2024年

作者： Ou, Ni Chen, Zhuo Luo, Shan The State Key Laboratory of Intelligent Control and Decision of Complex Systems Beijing Institute of Technology Beijing100081 China The Robot Perception Lab. Centre for Robotics Research Department of Engineering King’s College London LondonWC2R 2LS United Kingdom

Optical tactile sensors play a pivotal role in robot perception and manipulation tasks. The membrane of these sensors can be painted with markers or remain markerless, enabling them to function in either marker or markerless mode. However, this uni-modal selection means the sensor is only suitable for either manipulation or perception tasks. While markers are vital for manipulation, they can also obstruct the camera, thereby impeding perception. The dilemma of selecting between marker and markerless modes presents a significant obstacle. To address this issue, we propose a novel mode-switchable optical tactile sensing approach that facilitates transitions between the two modes. The marker-to-markerless transition is achieved through a generative model, whereas its inverse transition is realized using a sparsely supervised regressive model. Our approach allows a single-mode optical sensor to operate effectively in both marker and markerless modes without the need for additional hardware, making it well-suited for both perception and manipulation tasks. Extensive experiments validate the effectiveness of our method. For perception tasks, our approach decreases the number of categories that include misclassified samples by 2 and improves contact area segmentation IoU by 3.53%. For manipulation tasks, our method attains a high success rate of 92.59% in slip detection. Code, dataset and demo videos are availab.e at the project website https://***/Marker-Markerless-Transition/ Copyright © 2024, The Authors. All rights reserved.

关键词： Tactile sensors

来源：评论

学校读者我要写书评

暂无评论

Universal Process Migration Solution of MAGICAL for Analog IC Layout Automation

Universal Process Migration Solution of MAGICAL for Analog I...

引用

China Semiconductor technology International Conference (CSTIC)

作者： Yufeng Wei Yifan Xu Keren Zhu Ye Lu State Key Lab. of Integrated Chips and Systems School of Information Science and Technology Fudan University Shanghai China Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong China

ISBN: (数字)9798350362190

ISBN: (纸本)9798350362206

Analog IC layout automation has been a hot topic in the field in recent years[1]. An open-source software MAGICAL developed by the University of Texas team is one of the most effective means to achieve this goal[2]. However, so far MAGICAL is only able to support one process library currently, i.e., TSMC 40nm, and this imposes limitations of its practical application. This work proposes a solution for the technology migration of MAGICAL, and this greatly broadens its application scenarios. Firstly, the MAGICAL's device generator is rewritten to take in the general design rule information of different technology nodes. Second, the parameters of a given technology's tech file are extracted by customerized Parameter Extractor. Finally, the extracted parameters are set into the device generator of MAGICAL as well as the LEF file used for routing. As demonstrations, layouts of multiple circuits such as inverter, Foleded Cascode, Nbooster and Pbooster are generated sucessfully using this new flow at different technology nodesincluding but not limited to TSMC 65nm and TSMC 130nm. In addition, all GDS files generated with this method are DRC-clean and LVS-clean. This work paves the way for MAGICAL to be used in design automation at various technology nodes.

关键词： Integrated circuits Automation Design automation Layout Routing Generators Libraries

来源：评论

学校读者我要写书评

暂无评论

Compacting Binary Neural Networks by Sparse Kernel Selection

Compacting Binary Neural Networks by Sparse Kernel Selection

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Yikai Wang Wenbing Huang Yinpeng Dong Fuchun Sun Anbang Yao Department of Computer Science and Technology State Key Lab on Intelligent Technology and Systems BNRist Center Tsinghua University Gaoling School of Artificial Intelligence Renmin University of China RealAI Intel Labs China

Binary Neural Network (BNN) represents convolution weights with 1-bit values, which enhances the efficiency of storage and computation. This paper is motivated by a previously revealed phenomenon that the binary kernels in successful BNNs are nearly power-law distributed: their values are mostly clustered into a small number of codewords. This phenomenon encourages us to compact typical BNNs and obtain further close performance through learning non-repetitive kernels within a binary kernel subspace. Specifically, we regard the binarization process as kernel grouping in terms of a binary codebook, and our task lies in learning to select a smaller subset of codewords from the full code-book. We then leverage the Gumbel-Sinkhorn technique to approximate the codeword selection process, and develop the Permutation Straight-Through Estimator (PSTE) that is able to not only optimize the selection process end-to-end but also maintain the non-repetitive occupancy of selected codewords. Experiments verify that our method reduces both the model size and bit-wise computational costs, and achieves accuracy improvements compared with state-of-the-art BNNs under comparable budgets.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：