检索结果-内蒙古大学图书馆

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zou, Yuanbin Huang, Ziyun Xu, Jinhui Wang, Jianxin Feng, Qilong School of Computer Science and Engineering Central South University Changsha410083 China Xiangjiang Laboratory Changsha410205 China Department of Computer Science and Software Engineering Penn State Erie The Behrend College United States Department of Computer Science and Engineering State University of New York BuffaloNY United States The Hunan Provincial Key Lab of Bioinformatics Central South University Changsha410083 China

The Column Subset Selection (CSS) problem has been widely studied in dimensionality reduction and feature selection. The goal of the CSS problem is to output a submatrix S, consisting of k columns from an n × d input matrix A that minimizes the residual error A − SS†A2F, where S† is the Moore-Penrose inverse matrix of S. Many previous approximation algorithms have non-linear running times in both n and d, while the existing linear-time algorithms have a relatively larger approximation ratios. Additionally, the local search algorithms in existing results for solving the CSS problem are heuristic. To achieve linear running time while maintaining better approximation using a local search strategy, we propose a local search-based approximation algorithm for the CSS problem with exactly k columns selected. A key challenge in achieving linear running time with the local search strategy is how to avoid exhaustive enumerations of candidate columns for constructing swap pairs in each local search step. To address this issue, we propose a two-step mixed sampling method that reduces the number of enumerations for swap pair construction from O(dk) to k in linear time. Although the two-step mixed sampling method reduces the search space of local search strategy, bounding the residual error after swaps is a non-trivial task. To estimate the changes in residual error after swaps, we propose a matched swap pair construction method to bound the approximation loss, ensuring a constant probability of loss reduction in each local search step. In expectation, these techniques enable us to obtain the local search algorithm for the CSS problem with theoretical guarantees, where a 53(k + 1)-approximate solution can be obtained in linear running time O(ndk4 log k). Empirical experiments show that our proposed algorithm achieves better quality and time compared to previous algorithms on both small and large datasets. Moreover, it is at least 10 times faster than state-of-the-art algorithms a

关键词：

来源：评论

学校读者我要写书评

暂无评论

Near-Linear Time Approximation Algorithms for k-means with Outliers 41

Near-Linear Time Approximation Algorithms for k-means with O...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Huang, Junyu Feng, Qilong Huang, Ziyun Xu, Jinhui Wang, Jianxin School of Computer Science and Engineering Central South University Changsha410083 China Xiangjiang Laboratory Changsha410205 China Department of Computer Science and Software Engineering Penn State Erie The Behrend College United States Department of Computer Science and Engineering State University of New York BuffaloNY United States The Hunan Provincial Key Lab of Bioinformatics Central South University Changsha410083 China

The k-means with outliers problem is one of the most extensively studied clustering problems in the field of machine learning, where the goal is to discard up to z outliers and identify a minimum k-means clustering on the remaining data points. Most previous results for this problem have running time dependent on the aspect ratio ∆ (the ratio between the maximum and the minimum pairwise distances) to achieve fast approximations. To address the issue of aspect ratio dependency on the running time, we propose sampling-based algorithms with almost linear running time in the data size, where a crucial component of our approach is an algorithm called Fast-Sampling. Fast-Sampling algorithm can find inliers that well approximate the optimal clustering centers without relying on a guess for the optimal clustering costs, where a 4-approximate solution can be obtained in time O(ndk log log n/∊2) with O(k/∊) centers opened and (1 + ∊)z outliers discarded. To reduce the number of centers opened, we propose a center reduction algorithm, where an O(1/∊)-approximate solution can be obtained in time O(ndk log log n/∊2 + dpoly(k, 1/∊) log(n∆)) with (1 + ∊)z outliers discarded and exactly k centers opened. Empirical experiments suggest that our proposed sampling-based algorithms outperform state-of-the-art algorithms for the k-means with outliers problem. Copyright 2024 by the author(s)

关键词： K-means clustering

来源：评论

学校读者我要写书评

暂无评论

medIKAL: Integrating Knowledge Graphs as Assistants of LLMs for Enhanced Clinical Diagnosis on EMRs

arXiv

引用

arXiv 2024年

作者： Jia, Mingyi Duan, Junwen Song, Yan Wang, Jianxin Hunan Provincial Key Lab on Bioinformatics School of Computer Science and Engineering Central South University China University of Science and Technology of China China

Electronic Medical Records (EMRs), while integral to modern healthcare, present challenges for clinical reasoning and diagnosis due to their complexity and information redundancy. To address this, we proposed medIKAL (Integrating Knowledge Graphs as Assistants of LLMs), a framework that combines Large Language Models (LLMs) with knowledge graphs (KGs) to enhance diagnostic capabilities. medIKAL assigns weighted importance to entities in medical records based on their type, enabling precise localization of candidate diseases within KGs. It innovatively employs a residual network-like approach, allowing initial diagnosis by LLMs to be merged into KG search results. Through a path-based reranking algorithm and a fill-in-the-blank style prompt template, it further refined the diagnostic process. We validated medIKAL’s effectiveness through extensive experiments on a newly introduced open-sourced Chinese EMR dataset, demonstrating its potential to improve clinical diagnosis in real-world settings. The code and dataset are publicly available at https://***/CSU-NLP-Group/mediKAL. Copyright © 2024, The Authors. All rights reserved.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

On Fixed-Order Book Thickness Parameterized by the Pathwidth of the Vertex Ordering 14th

On Fixed-Order Book Thickness Parameterized by the Pathwidth...

引用

14th International Conference on Algorithmic Aspects in Information and Management, AAIM 2020

作者： Liu, Yunlong Chen, Jie Huang, Jingui Wang, Jianxin Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing Hunan Normal University Changsha410081 China School of Computer Science and Engineering Central South University Changsha410083 China

ISBN: (纸本)9783030576011

Given a graph (Formula Presented) and a fixed linear order (Formula Presented) of V, the problem fixed-order book thickness asks whether there is a page assignment (Formula Presented) such that (Formula Presented) is a k-page book embedding of G. Recently, Bhore et al.(GD2019) presented an algorithm parameterized by the pathwidth of the vertex ordering (denoted by (Formula Presented)). In this paper, we first re-analyze the running time for Bhore et al.’s algorithm, and prove a bound of (Formula Presented) improving on Bhore et al.’s bound of (Formula Presented). We further show that this parameterized problem does not admit a polynomial kernel unless NP (Formula Presented) coNP/poly. Finally, we show that the general fixed-order book thickness problem, in which a budget of at most c crossings over all pages was given, admits an algorithm running in time (Formula Presented). © 2020, Springer Nature Switzerland AG.

关键词： Budget control

来源：评论

学校读者我要写书评

暂无评论

MHLAT: Multi-Hop label-Wise Attention Model for Automatic ICD Coding

MHLAT: Multi-Hop Label-Wise Attention Model for Automatic IC...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Junwen Duan Han Jiang Ying Yu School of Computer Science and Engineering Central South University Changsha Hunan China Hunan Provincial Key Lab on Bioinformatics Central South University Changsha Hunan China

International Classification of Diseases (ICD) coding is the task of assigning ICD diagnosis codes to clinical notes. This can be challenging given the large quantity of labels (nearly 9,000) and lengthy texts (up to 8,000 tokens). However, unlike the single-pass reading process in previous works, humans tend to read the text and label definitions again to get more confident answers. Moreover, although pretrained language models have been used to address these problems, they suffer from huge memory usage. To address the above problems, we propose a simple but effective model called the Multi-Hop label-wise ATtention (MHLAT), in which multi-hop label-wise attention is deployed to get more precise and informative representations. Extensive experiments on three benchmark MIMIC datasets indicate that our method achieves significantly better or competitive performance on all seven metrics, with much fewer parameters to optimize.

关键词： Measurement Codes Speech coding MIMICs Benchmark testing Signal processing Acoustics

来源：评论

学校读者我要写书评

暂无评论

MHLAT: MULTI-HOP labEL-WISE ATTENTION MODEL FOR AUTOMATIC ICD CODING

arXiv

引用

arXiv 2023年

作者： Duan, Junwen Jiang, Han Yu, Ying School of Computer Science and Engineering Central South University Hunan Changsha China Hunan Provincial Key Lab on Bioinformatics Central South University Hunan Changsha China

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

GraphIsoFun: a graph neural network based approach for splice isoform function prediction

GraphIsoFun: a graph neural network based approach for splic...

引用

IEEE International Conference on bioinformatics and Biomedicine (BIBM)

作者： Shuo Zhang Changhuo Yang Hong-Dong Li Jianxin Wang School of Computer Science and Engineering Central South University Changsha China Hunan Hunan Provincial Key Lab on Bioinformatics Central South University Changsha China Hunan

ISBN: (纸本)9781665429825

Alternative splicing has a significant impact in the process of gene regulation and it greatly raises the complexity of protein species and functions. At present, the gene function has been extensively studied, and there are multiple public databases of gene function annotations. Due to the high cost and the large number of isoforms, it is currently impossible to conduct extensive experimental analysis on isoform functions. Functions at the level of splicing isoforms remain largely unknown. Computational prediction provides an alternative to functional annotation of splice isoforms. Here we present GraphIsoFun, a novel computational approach to predict isoform function by integrating both gene and isoform-level expression information with graph neural network. It first constructs a heterogeneous co-expression network with both genes and isoforms being nodes. This network integrates the rich association between isoforms and genes, and provides a bridge for connecting isoform and gene functions. Then GraphIsoFun uses the graph neural network to mine the information of the co-expression network and obtains the final prediction score of isoforms for each function. The experiments on multiple datasets show that GraphIsoFun surpasses existing methods based on different evaluation indicators.

关键词： Proteins Protein engineering Costs Annotations Databases Splicing Conferences

来源：评论

学校读者我要写书评

暂无评论

label distribution learning based on ensemble neural networks 25th

Label distribution learning based on ensemble neural network...

引用

25th International Conference on Neural Information Processing, ICONIP 2018

作者： Zhai, Yansheng Dai, Jianhua Shi, Hong School of Computer Science and Technology Tianjin University Tianjin300350 China Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing College of Information Science and Engineering Hunan Normal University Changsha410081 China

ISBN: (纸本)9783030041816

label distribution learning (LDL), as an extension of multi-label learning, is a new arising machine learning technique to deal with label ambiguity problems. The maximum entropy model is commonly used in label distribution learning. However, it does not consider the correlation between the labels and is not suitable for nonlinear relationships, and the prediction performance is also limited. In this paper, we propose a label distribution learning algorithm based on ensemble neural networks. The algorithm trains neural networks with preferences using training sets with different label sets to construct base learners, and combines the base learners with the weights, which is learned by the combined learner to obtain the final learning results. Experimental results show that the proposed algorithm is effective for label distribution data. © Springer Nature Switzerland AG 2018.

关键词： Learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Dynamic maintenance of decision rules for decision attribute values’ changing 25th

Dynamic maintenance of decision rules for decision attribute...

引用

25th International Conference on Neural Information Processing, ICONIP 2018

作者： Wang, Yingyao Dai, Jianhua Shi, Hong School of Computer Science and Technology Tianjin University Tianjin300350 China Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing College of Information Science and Engineering Hunan Normal University Changsha410081 China

ISBN: (纸本)9783030041816

Rule induction method based on rough set theory (RST) which can generate a minimal set of decision rules by using attribute reduction and approximations has received much attention recently. In real-life, the variation of objects, attributes and attributes’ values affects reducts and approximations, e.g., the coarsening and refining of attributes’ values. The goal of this paper is dynamic maintenance of decision rules for decision attribute values’ coarsening and refining. Two incremental rough-set based methods are proposed to deal with this issue by updating assignment discernibility matrix dynamically without recomputing the reducts from the beginning, which increases the efficiency. © Springer Nature Switzerland AG 2018.

关键词： Rough set theory

来源：评论

学校读者我要写书评

暂无评论

Local Adaptive U-net for Medical Image Segmentation

Local Adaptive U-net for Medical Image Segmentation

引用

IEEE International Conference on bioinformatics and Biomedicine (BIBM)

作者： Ning Liu Liangliang Liu Jianxin Wang Hunan Provincial Key Lab on Bioinformatics School of Computer Science and Engineering Central South University Changsha China Pingdingshan University Pingdingshan China

ISBN: (数字)9781728162157

ISBN: (纸本)9781728162164

Medical image segmentation is the primary measure of medical image analysis. With the development of deep learning, U-net based approaches have presented for different medical image segmentation tasks. However, the pooling and the simple convolution operation for deep feature maps in the U-shaped network would lead to the coarse segmentation result. In this paper, we design a local adaptive U-net (LA U-net) for medical image segmentation. There are two major modules: the Local Adaptive Module (LAM) and Multi-scale Convolution Module (MCM) in the network. The LAM get more feature maps from each down-sampling process. The MCM capture more global information for the encoding path. To validate the proposed network's performance, we verify it on two datasets: DRIVE dataset, and ISIC 2018 dataset; the results show that LA U-net achieves superior performance on two datasets.

关键词： Image segmentation Biomedical imaging Convolution Feature extraction Task analysis Lesions Sensitivity

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：