检索结果-内蒙古大学图书馆

28th Conference on Uncertainty in Artificial Intelligence, UAI 2012

作者： Amizadeh, Saeed Thiesson, Bo Hauskrecht, Milos Intelligent Systems Program University of Pittsburgh Pittsburgh PA 15213 United States Microsoft Research Redmond WA 98052 United States Department of Computer Science University of Pittsburgh Pittsburgh PA 15213 United States

ISBN: (纸本)9780974903989

In recent years, non-parametric methods utilizing random walks on graphs have been used to solve a wide range of machine learning problems, but in their simplest form they do not scale well due to the quadratic complexity. In this paper, a new dual-tree based variational approach for approximating the transition matrix and efficiently performing the random walk is proposed. The approach exploits a connection between kernel density estimation, mixture modeling, and random walk on graphs in an optimization of the transition matrix for the data graph that ties together edge transitions probabilities that are similar. Compared to the de facto standard approximation method based on k-nearestneighbors, we demonstrate order of magnitudes speedup without sacrificing accuracy for Label Propagation tasks on benchmark data sets in semi-supervised learning.

关键词： Random processes

来源：评论

学校读者我要写书评

暂无评论

Decomposition and Reorganization of Phonetic Information for Speaker Embedding Learning

引用

IEEE/ACM Transactions on Audio Speech and Language Processing 2023年 31卷 1745-1757页

作者： Hong, Qian-Bei Wu, Chung-Hsien Wang, Hsin-Min National Cheng Kung University and Academia Sinica Graduate Program of Multimedia Systems and Intelligent Computing Tainan701 Taiwan Academia Sinica Taipei115 Taiwan National Cheng Kung University Department of Computer Science and Information Engineering Tainan701401 Taiwan Academia Sinica Institute of Information Science Taipei115 Taiwan

Speech content is closely related to the stability of speaker embeddings in speaker verification tasks. In this paper, we propose a novel architecture based on self-constraint learning (SCL) and reconstruction task (RT) to remove the influence of phonetic information on speaker embedding generation. First, SCL is used to reduce the divergence of frame-level features, which can avoid ambiguity between the resulting embeddings of the two utterances being compared. Second, RT is used to further remove phonetic information in frame-level layers, focusing on speaker-discriminative feature transformation. In our experiments, the speaker embedding models were trained on the VoxCeleb2 dataset and evaluated on the VoxCeleb1, Librispeech, SITW and VoxMovies datasets. Experimental results on VoxCeleb1 show that the proposed DROP-TDNN system reduced the EER by 7.5%, compared to the state-of-the-art ECAPA-TDNN system. Furthermore, the proposed DROP-TDNN system also outperformed the ECAPA-TDNN system in the experiments on SITW, Librispeech and VoxMovies under cross-dataset conditions. In the experiments on SITW, the proposed system reduced the EER by 3.4% compared to the ECAPA-TDNN system. In the experiments on Librispeech, the proposed system demonstrated the advantage of removing phonetic information under the clean speech condition, with a significant reduction of 25.5% in EER compared to the ECAPA-TDNN system. In the experiments on VoxMovies, the proposed system reduced the EER by up to 7.9% compared to the ECAPA-TDNN system under different pronunciation and background conditions. © 2014 IEEE.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

ASSIST: Adaptive social support for information space traversal 07

ASSIST: Adaptive social support for information space traver...

引用

Hypertext 2007: 18th ACM Conference on Hypertext and Hypermedia, HT'07

作者： Farzan, Rosta Coyle, Maurice Freyne, Jill Brusilovsky, Peter Smyth, Barry University of Pittsburgh Intelligent Systems Program Pittsburgh PA 15260 United States University College Dublin School of Computer Science and Informatics Belfield Dublin 4 Ireland University of Pittsburgh School of Information Science Pittsburgh PA 15260 United States

ISBN: (纸本)1595938206

Finding relevant information in a hyperspace has been a much studied problem for many years. With the emergence of so called Web 2.0 technologies we have seen the use of social systems for retrieval tasks increasing dramatically. Each system collects and exploits its own pool of community wisdom for the benefit of its users. In this paper we suggest a form of retrieval which exploits the pools of wisdom of multiple social technologies, specifically social search and social navigation. The paper details the added user benefits of merging several sources of social wisdom. We present details of the ASSIST engine developed to integrate social support mechanisms for the users of information repositories. The goal of this paper is to present the main features of the integrated community-based personalization engine that we have developed in order to improve retrieval in the hyperspace of information resources. It also reports the results of an empirical study of this technology. Copyright 2007 ACM.

关键词： World Wide Web

来源：评论

学校读者我要写书评

暂无评论

Argument graph classification via genetic programming and C4.5

Argument graph classification via genetic programming and C4...

引用

1st International Conference on Educational Data Mining, EDM 2008

作者： Lynch, Collin Ashley, Kevin D. Pinkwart, Niels Aleven, Vincent Intelligent Systems Program University of Pittsburgh United States School of Law University of Pittsburgh United States Department of Informatics Clausthal University of Technology Germany HCII School of Computer Science Carnegie Mellon University United States

ISBN: (纸本)9780615306292

In well-defined domains there exist well-accepted criteria for detecting good and bad student solutions. Many ITS implement these criteria characterize solutions and to give immediate feedback. While this has been shown to promote learning, it is not always possible in ill-defined domains that typically lack well-accepted criteria. In this paper we report on the induction of classification rules for student solutions in an ill-defined domain. 1 We compare the viability of classifications using statistical measures with classification trees induced via C4.5 and Genetic programming.

关键词： Genetic programming

来源：评论

学校读者我要写书评

暂无评论

Modeling dynamic systems with memory: What is the right time-order?

Modeling dynamic systems with memory: What is the right time...

引用

8th Bayesian Modeling ApplicationsWorkshop, BMAW 2011

作者： Lupínska-Dubicka, Anna Druzdzel, Marek J. Faculty of Computer Science Bialystok University of Technology Wiejska 45A 15-351 Bialystok Poland Decision Systems Laboratory School of Information Sciences and Intelligent Systems Program University of Pittsburgh Pittsburgh PA 15260 United States

Most practical uses of Dynamic Bayesian Networks (DBNs) involve temporal inuences of the first order, i.e., inuences between neighboring time steps. This choice is a convenient approximation inuenced by the existence of efficient algorithms for first order models and limitations of available tools. We focus on the question whether constructing higher time-order models is worth the effort when the underlying system's memory goes beyond the current state. We present the results of an experiment with a series of DBN models monitoring woman's monthly cycle. We show that higher order models are significantly more accurate. However, we have also observed overfitting and a resulting decrease in accuracy when the time order chosen is too high.

关键词： Bayesian networks

来源：评论

学校读者我要写书评

暂无评论

The estrous cycle modulates hippocampal spine dynamics, dendritic processing, and spatial coding

引用

Neuron 2025年

作者： Wolcott, Nora S. Redman, William T. Karpinska, Marie Jacobs, Emily G. Goard, Michael J. Department of Molecular Cellular and Developmental Biology University of California Santa Barbara Santa Barbara 93106 CA United States Interdepartmental Graduate Program in Dynamical Neuroscience University of California Santa Barbara Santa Barbara 93106 CA United States Intelligent Systems Center Johns Hopkins University Applied Physics Laboratory Laurel 20723 MD United States Department of Computer Science University of California Santa Barbara Santa Barbara 93106 CA United States Department of Psychological & Brain Sciences University of California Santa Barbara Santa Barbara 93106 CA United States Ann S. Bowers Women's Brain Health Initiative University of California Santa Barbara Santa Barbara 93106 CA United States Neuroscience Research Institute University of California Santa Barbara Santa Barbara 93106 CA United States

Histological evidence suggests that the estrous cycle exerts a powerful influence on CA1 neurons in the mammalian hippocampus. Decades have passed since this landmark observation, yet how the estrous cycle shapes dendritic spine dynamics and hippocampal spatial coding in vivo remains a mystery. Here, we used a custom hippocampal microperiscope and two-photon calcium imaging to track CA1 pyramidal neurons in female mice across multiple cycles. Estrous cycle stage had a potent effect on spine dynamics, with spine density peaking during proestrus when estradiol levels are highest. These morphological changes coincided with greater somatodendritic coupling and increased infiltration of back-propagating action potentials into the apical dendrite. Finally, tracking CA1 response properties during navigation revealed greater place field stability during proestrus, evident at both the single-cell and population levels. These findings demonstrate that the estrous cycle drives large-scale structural and functional plasticity in hippocampal neurons essential for learning and memory. © 2025 The Author(s)

关键词： dendritic integration estradiol estrous cycle hippocampus multiphoton imaging neuroendocrine place cells spine morphology

来源：评论

学校读者我要写书评

暂无评论

Making large Cox's proportional hazard models tractable in Bayesian networks 8

Making large Cox's proportional hazard models tractable in B...

引用

8th International Conference on Probabilistic Graphical Models, PGM 2016

作者： Kraisangka, Jidapa Druzdzel, Marek J. Decision System Laboratory School of Information Sciences Intelligent Systems Program University of Pittsburgh PittsburghPA15260 United States Faculty of Computer Science Bialystok University of Technology Wiejska 45A Bialystok15-351 Poland

Cox's proportional hazard (CPH) model is a statistical technique that captures the interaction between a set of risk factors and an effect variable. While the CPH model is popular in survival analysis, Bayesian networks offer an attractive alternative that is intuitive, general, theoretically sound, and avoids CPH model's restrictive assumptions. Existing CPH models are a great source of existing knowledge that can be reused in Bayesian networks. The main problem with applying Bayesian networks to survival analysis is their exponential growth in complexity as the number of risk factors increases. It is not uncommon to see complex CPH models with as many as 20 risk factors. Our paper focuses on making large survival analysis models derived from the CPH model tractable in Bayesian networks. We evaluate the effect of two complexity reduction techniques: (1) parent divorcing, and (2) removing less important risk factors based on the accuracy of the resulting models. 2016 Microtome Publishing. All rights reserved.

关键词： Bayesian networks

来源：评论

学校读者我要写书评

暂无评论

Learning discrete Bayesian network parameters from continuous data streams: What is the best strategy?

Learning discrete Bayesian network parameters from continuou...

引用

作者： Ratnapinda, Parot Druzdzel, Marek J. Decision Systems Laboratory School of Information Sciences and Intelligent Systems Program University of Pittsburgh PittsburghPA15260 United States Faculty of Science Information Technology Division Maejo University Chiang Mai50290 Thailand Faculty of Computer Science Bialystok University of Technology Wiejska 45A Bialystok15-351 Poland

We compare three approaches to learning numerical parameters of discrete Bayesian networks from continuous data streams: (1) the EM algorithm applied to all data, (2) the EM algorithm applied to data increments, and (3) the online EM algorithm. Our results show that learning from all data at each step, whenever feasible, leads to the highest parameter accuracy and model classification accuracy. When facing computational limitations, incremental learning approaches are a reasonable alternative. While the differences in speed between incremental algorithms are not large (online EM is slightly slower), for all but small data sets online EM tends to be more accurate than incremental EM. ©2015 Elsevier B.V. All rights reserved.

关键词： Bayesian networks

来源：评论

学校读者我要写书评

暂无评论

Comparing user simulation models for dialog strategy learning

Comparing user simulation models for dialog strategy learnin...

引用

2007 Human Language Technology Conference of the North American Chapter of the Association of Computational Linguistics, NAACL-HLT 2007

作者： Ai, Hua Tetreault, Joel R. Litman, Diane J. University of Pittsburgh Intelligent Systems Program PittsburghPA15260 United States University of Pittsburgh LRDC PittsburghPA15260 United States University of Pittsburgh Dept. of Computer Science LRDC PittsburghPA15260 United States

This paper explores what kind of user simulation model is suitable for developing a training corpus for using Markov Decision Processes (MDPs) to automatically learn dialog strategies. Our results suggest that with sparse training data, a model that aims to randomly explore more dialog state spaces with certain constraints actually performs at the same or better than a more complex model that simulates realistic user behaviors in a statistical way. © 2007 Association for Computational Linguistics.

关键词： Markov processes

来源：评论

学校读者我要写书评

暂无评论

OpinionFinder: A system for subjectivity analysis

OpinionFinder: A system for subjectivity analysis

引用

2005 Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing, HLT/EMNLP 2005

作者： Wilson, Theresa Hoffmann, Paul Somasundaran, Swapna Kessler, Jason Wiebe, Janyce Choi, Yejin Cardie, Claire Riloff, Ellen Patwardhan, Siddharth Intelligent Systems Program University of Pittsburgh PittsburghPA15260 United States Department of Computer Science University of Pittsburgh PittsburghPA15260 United States Department of Computer Science Cornell University IthacaNY14853 United States School of Computing University of Utah Salt Lake CityUT84112 United States

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：