检索结果-内蒙古大学图书馆

Proceedings of the 37th International Conference on Neural Information processing Systems

作者： Lingchen Meng Xiyang Dai Jianwei Yang Dongdong Chen Yinpeng Chen Mengchen Liu Yi-Ling Chen Zuxuan Wu Lu Yuan Yu-Gang Jiang Shanghai Key Lab of Intell. Info. Processing School of CS Fudan University and Shanghai Collaborative Innovation Center of Intelligent Visual Computing Microsoft

Long-tailed object detection (LTOD) aims to handle the extreme data imbalance in real-world datasets, where many tail classes have scarce instances. One popular strategy is to explore extra data with image-level labels, yet it produces limited results due to (1) semantic ambiguity—an image-level label only captures a salient part of the image, ignoring the remaining rich semantics within the image; and (2) location sensitivity—the label highly depends on the locations and crops of the original image, which may change after data transformations like random cropping. To remedy this, we propose RichSem, a simple but effective method, which is robust to learn rich semantics from coarse locations without the need of accurate bounding boxes. RichSem leverages rich semantics from images, which are then served as additional "soft supervision" for training detectors. Specifically, we add a semantic branch to our detector to learn these soft semantics and enhance feature representations for long-tailed object detection. The semantic branch is only used for training and is removed during inference. RichSem achieves consistent improvements on both overall and rare-category of LVIS under different backbones and detectors. Our method achieves state-of-the-art performance without requiring complex training and testing procedures. Moreover, we show the effectiveness of our method on other long-tailed datasets with additional experiments. Code is available at https://***/MengLcool/RichSem.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MSMDFusion: Fusing LiDAR and Camera at Multiple Scales with Multi-Depth Seeds for 3D Object Detection

MSMDFusion: Fusing LiDAR and Camera at Multiple Scales with ...

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Yang Jiao Zequn Jie Shaoxiang Chen Jingjing Chen Lin Ma Yu-Gang Jiang Shanghai Key Lab of Intell. Info. Processing School of CS Fudan University Shanghai Collaborative Innovation Center of Intelligent Visual Computing Meituan

Fusing LiDAR and camera information is essential for accurate and reliable 3D object detection in autonomous driving systems. This is challenging due to the difficulty of combining multi-granularity geometric and semantic features from two drastically different modalities. Recent approaches aim at exploring the semantic densities of camera features through lifting points in 2D camera images (referred to as “seeds”) into 3D space, and then incorporate 2D semantics via cross-modal interaction or fusion techniques. However, depth information is under-investigated in these approaches when lifting points into 3D space, thus 2D semantics can not be reliably fused with 3D points. Moreover, their multi-modal fusion strategy, which is implemented as concatenation or attention, either can not effectively fuse 2D and 3D information or is unable to perform fine-grained interactions in the voxel space. To this end, we propose a novel framework with better utilization of the depth information and fine-grained cross-modal interaction between LiDAR and camera, which consists of two important components. First, a Multi-Depth Unprojection (MDU) method is used to enhance the depth quality of the lifted points at each interaction level. Second, a Gated Modality-Aware Convolution (GMA-Conv) block is applied to modulate voxels involved with the camera modality in a fine-grained manner and then aggregate multi-modal features into a unified space. Together they provide the detection head with more comprehensive features from LiDAR and camera. On the nuScenes test benchmark, our proposed method, abbreviated as MSMD-Fusion, achieves state-of-the-art results on both 3D object detection and tracking tasks without using test-time-augmentation and ensemble techniques. The code is available at https://***/SxJyJay/MSMDFusion.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration

arXiv

引用

arXiv 2024年

作者： Wen, Zhuofan Gui, Shangtong Feng, Yang Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences China State Key Lab of Processors Institute of Computing Technology Chinese Academy of Sciences China Key Laboratory of Al Safety Chinese Academy of Sciences China University of Chinese Academy of Sciences Beijing China

Inference acceleration of large language models (LLMs) has been put forward in many application scenarios and speculative decoding has shown its advantage in addressing inference acceleration. Speculative decoding usually introduces a draft model to assist the base LLM where the draft model produces drafts and the base LLM verifies the draft for acceptance or rejection. In this framework, the final inference speed is decided by the decoding speed of the draft model and the acceptance rate of the draft provided by the draft model. Currently the widely used draft models usually generate draft tokens for the next several positions in a non-autoregressive way without considering the correlations between draft tokens. Therefore, it has a high decoding speed but an unsatisfactory acceptance rate. In this paper, we focus on how to improve the performance of the draft model and aim to accelerate inference via a high acceptance rate. To this end, we propose a CTC-based draft model which strengthens the correlations between draft tokens during the draft phase, thereby generating higher-quality draft candidate sequences. Experiment results show that compared to strong baselines, the proposed method can achieve a higher acceptance rate and hence a faster inference speed. Copyright © 2024, The Authors. All rights reserved.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Adaptive Split-Fusion Transformer

Adaptive Split-Fusion Transformer

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Zixuan Su Jingjing Chen Lei Pang Chong-Wah Ngo Yu-Gang Jiang Shanghai Key Lab of Intelligent Information Processing School of Computer Science Fudan University Shanghai Collaborative Innovation Center of Intelligent Visual Computing City University of Hong Kong Singapore Management University

Neural networks for visual content understanding have recently evolved from convolutional ones to transformers. The prior (CNN) relies on small-windowed kernels to capture the regional clues, demonstrating solid local expressiveness. On the contrary, the latter (transformer) establishes long-range global connections between localities for holistic learning. Inspired by this complementary nature, there is a growing interest in designing hybrid models which utilize both techniques. Current hybrids merely replace convolutions as simple approximations of linear projection or juxtapose a convolution branch with attention without considering the importance of local/global modeling. To tackle this, we propose a new hybrid named Adaptive Split-Fusion Transformer (ASF-former) that treats convolutional and attention branches differently with adaptive weights. Specifically, an ASF-former encoder equally splits feature channels into half to fit dual-path inputs. Then, the outputs of the dual-path are fused with weights calculated from visual cues. We also design a compact convolutional path from a concern of efficiency. Extensive experiments on standard benchmarks show that our ASF-former outperforms its CNN, transformer, and hybrid counterparts in terms of accuracy (83.9% on ImageNet-1K), under similar conditions (12.9G MACs / 56.7M Params, without large-scale pre-training). The code is available at: https://***/szx503045266/ASF-former.

关键词：

来源：评论

学校读者我要写书评

暂无评论

DeepStack: Deeply Stacking Visual Tokens is Surprisingly Simple and Effective for LMMs

arXiv

引用

arXiv 2024年

作者： Meng, Lingchen Yang, Jianwei Tian, Rui Dai, Xiyang Wu, Zuxuan Gao, Jianfeng Jiang, Yu-Gang Shanghai Key Lab of Intell. Info. Processing School of CS Fudan University China Shanghai Collaborative Innovation Center of Intelligent Visual Computing China Microsoft Corporation United States

Most large multimodal models (LMMs) are implemented by feeding visual tokens as a sequence into the first layer of a large language model (LLM). The resulting architecture is simple but significantly increases computation and memory costs, as it has to handle a large number of additional tokens in its input layer. This paper presents a new architecture DeepStack for LMMs. Considering N layers in the language and vision transformer of LMMs, we stack the visual tokens into N groups and feed each group to its aligned transformer layer from bottom to top, as illustrated in Fig. 1. Surprisingly, this simple method greatly enhances the power of LMMs to model interactions among visual tokens across layers but with minimal additional cost. We apply DeepStack to both language and vision transformer in LMMs, and validate the effectiveness of DeepStack LMMs with extensive empirical results. Using the same context length, our DeepStack 7B and 13B parameters surpass their counterparts by 2.7 and 2.9 on average across 9 benchmarks, respectively. Using only one-fifth of the context length, DeepStack rivals closely to the counterparts Copyright © 2024, The Authors. All rights reserved.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

NetPrompt: Neural Network Prompting Enhances Event Extraction in Large Language Models

引用

IEEE Transactions on Big Data 2025年

作者： Mu, Lin Cheng, Yide Shen, Jun Zhang, Yiwen Zhong, Hong School of Computer Science and Technology Anhui University Anhui Hefei230601 China The Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education the Anhui Engineering Laboratory of IoT Security Technologies the School of Computer Science and Technology the Institute of Physical Science and Information Technology Anhui University Hefei230039 China

Event Extraction involves extracting event-related information such as event types and event arguments from context, which has long been tackled through well-designed neural networks or fine-tuned pre-trained language models. These approaches require substantial annotated data for tuning parameters and are resource-intensive. Recently, Prompting strategies with frozen parameters, such as Chain-of-Thought and Self-Consistency, have delivered success in NLP using LLMs by generating intermediate thought steps. However, they suffer from the challenge of error propagation and lack of interaction between different thoughts. In this paper, we propose Neural Network-based Prompting (NetPrompt), a novel network-structured prompting strategy for event extraction. The core idea behind NetPrompt is to imitate the excellent information integration capabilities of neural network structures. Specifically, we first decompose the event extraction problem into diverse intermediate subtasks, and each subtask is represented as a node in different layers of the network, the output of the nodes in the preceding layer is fed into the subsequent layer. Secondly, we propose pruning strategies to adapt the reasoning overhead to different problems. Finally, we have conducted extensive experiments on two widely used event extraction benchmarks to evaluate NetPrompt. The results demonstrated that NetPrompt significantly improved the event extraction performance compared to previous methods. ©2015 IEEE.

关键词： Neural network models

来源：评论

学校读者我要写书评

暂无评论

A dual-band antenna designed for 5G communication based on a self-decoupled antenna

A dual-band antenna designed for 5G communication based on a...

引用

International Applied Computational Electromagnetics Society Symposium (ACES)

作者： Rongzhen Fang Xianliang Wu Zhipeng Chen Mengying Dong Yunfeng Hu Yuanyuan Li Key Laboratory of Intelligent Computing and Signal Processing Anhui University Hefei China Department of Education of Anhui Province Anhui University Hefei China

In this paper, a dual-band antenna for 5G communication based on an antenna design with self-decoupling properties is *** antenna is composed of a self-decoupled antenna unit vertically placed on an 30×80 mm 2 ground plane and printed on 0.8 mm thick FR-4(εr = 4.4, tanδ = 0.02) substrate,and then added a pair of L-shaped coupling feeder structures on this *** shows that the antenna has two operating frequency bands of 3.3-4.2GHz and 4.8-5GHz, and has good transmission and isolation in the above two operating frequency ***,it also has the advantages of small size, self-decoupling,high isolation,simple structure and easy *** antenna can be used as 5G mobile phone communication antenna unit.

关键词： Couplings 5G mobile communication Transmitting antennas Dual band Mobile antennas Antenna feeds Computational electromagnetics

来源：评论

学校读者我要写书评

暂无评论

Electromagnetic scattering of two-dimensional rough cylinder based on statistical integral equation 2

Electromagnetic scattering of two-dimensional rough cylinder...

引用

2020 2nd International Conference on Artificial Intelligence Technologies and Application, ICAITA 2020

作者： Zhu, Mengnan Wang, Anqi Huang, Zhixiang Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei China

The study of electromagnetic scattering from Gaussian rough surface is of great significance in radar reconnaissance, target tracking and ocean remote sensing. The moment method (MOM) is a commonly used method with high accuracy. However, in the past research, Monte Carlo method is used to simulate the Gaussian rough surface. In this method, a large number of random simulation operations are carried out, and then a statistical convergence result is obtained by counting the results of all operations. In this paper, the stochastic integral equation method SIEM(Stochastic Integral Equation Method) is proposed to calculate the electromagnetic scattering of Gaussian rough surface. SIEM avoids the difficulty of high difficulty precise modeling and repeated calculation many times, and only single modeling and calculation are carried out for the target problem. In this paper, the electromagnetic scattering of two-dimensional cylinder with Gaussian rough surface is solved, and the effectiveness of SIEM is proved by comparing with mom. © 2020 Published under licence by IOP Publishing Ltd.

关键词： Method of moments

来源：评论

学校读者我要写书评

暂无评论

A RIS-Assisted Multiuser MIMO Communication Method Based on Deep Reinforcement Learning

A RIS-Assisted Multiuser MIMO Communication Method Based on ...

引用

IEEE International Conference on Communication and Information Systems (ICCIS)

作者： Changlin Tian Hui Zhi Kui Tie Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Anhui University School of Electronic and Information Engineering Anhui University Hefei China Ceyear Technologies (AnHui) CO. LTD Bengbu China

In this paper, a RIS-assisted multiuser MIMO communication method based on deep reinforcement learning (RMMC-DRL) is proposed for multiuser scenarios. Our objective is to find the optimal transmit beamforming matrix of BS and optimal phase shift matrix of reflective intelligent surface (RIS) to maximize the sum rate of multiuser, this problem is reduced into a constrained optimization problem. It is a non-convex optimization problem, so we solve it through deep reinforcement learning (DRL) and then use the results for communication. In the DRL, a deep deterministic policy gradient (DDPG) framework that can handle continuous states and actions is designed, reward is set as optimization goal, and the transmit beamforming matrix and the phase shift matrix of RIS are obtained through the interaction with environment. Unlike the alternating optimization (AO) method, which solve the transmit beamforming matrix and the RIS phase shift matrix alternatively, the RMMC-DRL can obtain both transmit beamforming matrix and RIS phase shift matrix simultaneously as the output of DRL. Simulation results show that RMMC-DRL can learn and improve its behavior by interacting with the environment. Compared with AO method, RMMC-DRL can obtain higher sum rate and lower computational complexity.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Expanding self-orthogonal codes over a ring Z4 to self-dual codes and unimodular lattices

arXiv

引用

arXiv 2024年

作者： Shi, Minjia Tao, Sihui Hong, Jihoon Kim, Jon-Lark The Key Laboratory of Intelligent Computing Signal Processing Ministry of Education School of Mathematical Sciences Anhui University Anhui Hefei230601 China State Key Laboratory of Integrated Service Networks Xidian University Xi’an710071 China The Department of Mathematics Sogang University Seoul Korea Republic of

Self-dual codes have been studied actively because they are connected with mathematical structures including block designs and lattices and have practical applications in quantum error-correcting codes and secret sharing schemes. Nevertheless, there has been less attention to construct self-dual codes from self-orthogonal codes with smaller dimensions. Hence, the main purpose of this paper is to propose a way to expand any self-orthogonal code over a ring Z4 to many self-dual codes over Z4. We show that all self-dual codes over Z4 of lengths 4 to 8 can be constructed this way. Furthermore, we have found five new self-dual codes over Z4 of lengths 27, 28, 29, 33, and 34 with the highest Euclidean weight 12. Moreover, using Construction A applied to our new Euclidean-optimal self-dual codes over Z4, we have constructed a new odd extremal unimodular lattice in dimension 34 whose kissing number was not previously known. © 2024, CC0.

关键词： Crystal lattices

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：