检索结果-内蒙古大学图书馆

Length of the state trace: A method for partitioning model complexity

JOURNAL OF MATHEMATICAL PSYCHOLOGY 2023年第1期113卷

作者： Ashby, F. Gregory Univ Calif Santa Barbara Santa Barbara CA 93106 USA

A novel and easy-to-compute measure is proposed that compares the relative contribution of each parameter of a mathematical model to the model's mathematical flexibility or complexity, with respect to accounting for the results of some specific experiment. When the data space is a two-dimensional plot of the type used in standard state-trace analysis, then the model complexity contributed by a single parameter equals the length of the state trace (LOST) that results when that parameter is varied and all other parameters are held constant. For the normal, equal-variance, signal-detection model, the average LOST when the response-criterion parameter XC is varied is about four times greater than the average LOST when the sensitivity parameter d ' is varied. As a result, applying the signal-detection model to random data almost always leads to the conclusion that all the points share the same value of d ' but were generated under different values of XC. Parameters that have non-monotonic effects on performance, such as the attention-weight parameter that is used in popular exemplar and prototype models of categorization, tend to have large LOSTs, and therefore contribute to model flexibility more than parameters that have monotonic effects on performance. Comparing LOSTs for exemplar and prototype models also leads to some deep new insights into the structure of both models.(c) 2023 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://***/licenses/by/4.0/).

关键词： model complexity State-trace analysis Signal-detection theory Generalized context model Prototype model

来源：评论

学校读者我要写书评

暂无评论

Comparing model complexity for glacio-hydrological simulation in the data-scarce Peruvian Andes

引用

JOURNAL OF HYDROLOGY-REGIONAL STUDIES 2021年 37卷

作者： Munoz, Randy Huggel, Christian Drenkhan, Fabian Vis, Marc Viviroli, Daniel Univ Zurich Dept Geog Winterthurerstr 190 CH-8057 Zurich Switzerland Imperial Coll London Dept Civil & Environm Engn London SW7 2AZ England Pontificia Univ Catolica Peru Dept Humanidades Lima Peru

Study region: Glaciated headwaters of the Vilcanota-Urubamba river basin, Southern Peru Study focus: A pivotal question is if robust hydrological simulation of streamflow in data-scarce and glaciated catchments can be achieved using parsimonious or more complex models. Therefore, a multi-model assessment of three glacio-hydrological models of different complexity was conducted thoroughly analyzing model performance, flow signatures and runoff components. New hydrological insights for the region: In data-scarce catchments, such as in the tropical Andes, parsimonious glacio-hydrological models can provide more robust results than complex models. While the overall performance of all models was reasonably good (R-2: 0.65-0.70, Nash-Sutcliffe: 0.65-0.73, Nash-Sutcliffe-ln: 0.73-0.78), with increasing data scarcity more complex models involve higher uncertainties. Furthermore, complex models require substantial understanding of the underpinning hydrological processes and a comprehensive calibration strategy to avoid apparently high model performance driven by inadequate assumptions. Based on these insights we present a framework for robust glacio-hydrological simulation under data scarcity. This stepwise approach includes, among others, a multi-model focus with a comprehensive assessment of flow signatures and runoff components. Future modeling needs to be further supported by alternative data collection strategies to substantially improve knowledge and process understanding. Therefore, the extension of sensor and station networks combined with the integration of co-produced knowledge represents a meaningful measure to robust decision-making for climate change adaptation and water management under high uncertainty.

关键词： Glacio-hydrological simulation Tropical Andes Shaman model Scarce data model complexity Multi-model comparison

来源：评论

学校读者我要写书评

暂无评论

On the role of deep learning model complexity in adversarial robustness for medical images

引用

BMC MEDICAL INFORMATICS AND DECISION MAKING 2022年第SUPPL 2期22卷 160-160页

作者： Rodriguez, David Nayak, Tapsya Chen, Yidong Krishnan, Ram Huang, Yufei Univ Texas San Antonio Dept Elect & Comp Engn San Antonio TX USA Univ Texas Hlth San Antonio Greehey Childrens Canc Res Inst San Antonio TX USA Univ Texas Hlth San Antonio Dept Populat Hlth Sci San Antonio TX USA Univ Pittsburgh Sch Med Dept Med UPMC Hillman Canc Ctr Pittsburgh PA 15213 USA

Background Deep learning (DL) models are highly vulnerable to adversarial attacks for medical image classification. An adversary could modify the input data in imperceptible ways such that a model could be tricked to predict, say, an image that actually exhibits malignant tumor to a prediction that it is benign. However, adversarial robustness of DL models for medical images is not adequately studied. DL in medicine is inundated with models of various complexity-particularly, very large models. In this work, we investigate the role of model complexity in adversarial settings. Results Consider a set of DL models that exhibit similar performances for a given task. These models are trained in the usual manner but are not trained to defend against adversarial attacks. We demonstrate that, among those models, simpler models of reduced complexity show a greater level of robustness against adversarial attacks than larger models that often tend to be used in medical applications. On the other hand, we also show that once those models undergo adversarial training, the adversarial trained medical image DL models exhibit a greater degree of robustness than the standard trained models for all model complexities. Conclusion The above result has a significant practical relevance. When medical practitioners lack the expertise or resources to defend against adversarial attacks, we recommend that they select the smallest of the models that exhibit adequate performance. Such a model would be naturally more robust to adversarial attacks than the larger models.

关键词： Adversarial attacks Perturbation Adversarial robustness Medical image classification model complexity

来源：评论

学校读者我要写书评

暂无评论

Evaluating three evapotranspiration estimates from model of different complexity over China using the ILAMB benchmarking system

引用

JOURNAL OF HYDROLOGY 2020年 590卷 125553-125553页

作者： Wu, Genan Cai, Xitian Keenan, Trevor F. Li, Shenggong Luo, Xiangzhong Fisher, Joshua B. Cao, Ruochen Li, Fa Purdy, Adam J. Zhao, Wei Sun, Xiaomin Hu, Zhongmin Chinese Acad Sci Synth Res Ctr Inst Geog Sci & Nat Resources Res Key Lab Ecosyst Network Observat & ModelingChin Beijing Peoples R China Univ Chinese Acad Sci Coll Resources & Environm Beijing Peoples R China Lawrence Berkeley Natl Lab Berkeley CA USA Univ Calif Berkeley Berkeley CA USA CALTECH Jet Prop Lab Pasadena CA USA South China Normal Univ Sch Geog Guangzhou Peoples R China Wuhan Univ State Key Lab Informat Engn Surveying Mapping & R Wuhan Peoples R China Southern Marine Sci & Engn Guangdong Lab Zhuhai Peoples R China

Land surface models range in complexity of terrestrial evapotranspiration, yet it is unknown how model complexity translates to accuracy of modeled evapotranspiration estimates. Here, we use the International Land model Benchmarking system to assess ET estimates from three models of varying complexity driven by the same forcing datasets: an earth system model, a terrestrial biosphere model, and a stand-alone ET model. The performance assessment includes both temporal and spatial evaluation, and different plant functional types across China. Our results indicate that the most complex model, an earth system model, performed best against the benchmarking datasets and metrics. Terrestrial biosphere model performed best in simulating inter-annual variability of ET, while earth system model performed best in simulating the seasonal cycle. The more complex models (earth system model and terrestrial biosphere model) perform better in forest, shrub and crop ecosystems, while the simpler model (stand-alone ET model) perform better in grass ecosystems. Our study demonstrates the impact of model complexity on ET estimates and highlights directions for future ET model improvements.

关键词： Benchmarking Evapotranspiration model model complexity

来源：评论

学校读者我要写书评

暂无评论

The effects of model and data complexity on predictions from species distributions models

引用

ECOLOGICAL modelLING 2016年 326卷 4-12页

作者： Garcia-Callejas, David Araujo, Miguel B. Univ Evora InBIO CIBIO P-7000 Evora Portugal Imperial Coll London Silwood Pk CampusBuckhurst Rd Ascot Berks England Natl Museum Nat Sci Calle Jose Gutierrez Abascal 2 Madrid 28006 Spain Univ Copenhagen Ctr Macroecol Evolut & Climate Nat Hist Museum Denmark Univ Pk 15 DK-2100 Copenhagen Denmark

How complex does a model need to be to provide useful predictions is a matter of continuous debate across environmental sciences. In the species distributions modelling literature, studies have demonstrated that more complex models tend to provide better fits. However, studies have also shown that predictive performance does not always increase with complexity. Testing of species distributions models is challenging because independent data for testing are often lacking, but a more general problem is that model complexity has never been formally described in such studies. Here, we systematically examine predictive performance of models against data and models of varying complexity. We introduce the concept of computational complexity, widely used in theoretical computer sciences, to quantify model complexity. In addition, complexity of species distributional data is characterized by their geometrical properties. Tests involved analysis of models' ability to predict virtual species distributions in the same region and the same time as used for training the models, and to project distributions in different times under climate change. Of the eight species distribution models analyzed five (Random Forest, boosted regression trees, generalized additive models, multivariate adaptive regression splines, MaxEnt) showed similar performance despite differences in computational complexity. The ability of models to forecast distributions under climate change was also not affected by model complexity. In contrast, geometrical characteristics of the data were related to model performance in several ways: complex datasets were consistently more difficult to model, and the complexity of the data was affected by the choice of predictors and the type of data analyzed. Given our definition of complexity, our study contradicts the widely held view that the complexity of species distributions models has significant effects in their predictive ability while findings support for previous o

关键词： Climate change Data complexity model complexity Species distributions models Transferability Virtual species

来源：评论

学校读者我要写书评

暂无评论

A study on Bonded Block model (BBM) complexity for simulation of laboratory-scale stress-strain behavior in granitic rocks

引用

COMPUTERS AND GEOTECHNICS 2020年 118卷 1页

作者： Sinha, Sankhaneel Walton, Gabriel Colorado Sch Mines Dept Geol & Geol Engn 1516 Illinois St Golden CO 80401 USA

The emergent macroscopic behaviors of Bonded Block models (BBM) are governed largely by the properties assigned to their three components blocks, contacts and zones (blocks are discretized using finite-difference zones). Over the years, different representations of these components, including elastic/inelastic zones and heterogeneous/homogenous blocks and contacts (corresponding to different mineral grains and associated mechanical properties), have been employed to simulate various aspects of rock behavior. However, there is a lack of understanding of the capabilities of these model representations with respect to their ability to replicate specific rock mechanical attributes. The goal of this study was to test a variety of model representations and evaluate their capabilities in terms of reproducing mechanical behaviors observed for a granitic rock type. It was found that inelastic zones were necessary to capture high confinement peak strengths while incorporation of heterogeneity was necessary to replicate the micro-cracking process. The heterogeneous, inelastic BBMs could match all the calibration targets and is identified as the best representation for modeling the full range of granitic rock behaviors. To help researchers select an appropriate BBM representation for modeling different aspects of rock behavior, a summary table outlining the capabilities of the different model representations is also provided.

关键词： Bonded block models Voronoi tessellation Rock fracturing model complexity

来源：评论

学校读者我要写书评

暂无评论

Property Inference Attacks on Convolutional Neural Networks: Influence and Implications of Target model's complexity 18

Property Inference Attacks on Convolutional Neural Networks:...

引用

18th International Conference on Security and Cryptography (SECRYPT)

作者： Parisot, Mathias Pejo, Balazs Spagnuelo, Dayana Univ Amsterdam Amsterdam Netherlands Budapest Univ Technol & Econ Dept Networked Syst & Serv CrySyS Lab Budapest Hungary Vrije Univ Amsterdam Amsterdam Netherlands

ISBN: (纸本)9789897585241

Machine learning models' goal is to make correct predictions for specific tasks by learning important properties and patterns from data. By doing so, there is a chance that the model learns properties that are unrelated to its primary task. Property Inference Attacks exploit this and aim to infer from a given model (i.e., the target model) properties about the training dataset seemingly unrelated to the model's primary goal. If the training data is sensitive, such an attack could lead to privacy leakage. In this paper, we investigate the influence of the target model's complexity on the accuracy of this type of attack, focusing on convolutional neural network classifiers. We perform attacks on models that are trained on facial images to predict whether someone's mouth is open. Our attacks' goal is to infer whether the training dataset is balanced gender-wise. Our findings reveal that the risk of a privacy breach is present independently of the target model's complexity: for all studied architectures, the attack's accuracy is clearly over the baseline.

关键词： Property Inference Attacks Convolutional Neural Networks model complexity

来源：评论

学校读者我要写书评

暂无评论

Decomposed Vector Combination-Based Low-complexity Behavioral model for Digital Predistortion of RF Transmitters

引用

IEEE TRANSACTIONS ON BROADCASTING 2024年第1期70卷 334-346页

作者： Han, Renlong Jiang, Chengye Yang, Guichen Zhang, Qianqian Liu, Falin Univ Sci & Technol China Dept Elect Engn & Informat Sci Hefei 230027 Peoples R China Chinese Acad Sci Key Lab Electromagnet Space Informat Hefei 230027 Peoples R China

In this paper, we present a novel behavioral modeling technique based on decomposed vector combination (DVC) for digital predistortion (DPD) of RF Transmitters. The basis function of the proposed DVC model consists of a piecewise function-based magnitude term and a linear phase-combination-based phase term. The novel DVC basis functions are still linear-in-parameters and go beyond the classical Volterra series. Compared with classical DPD models, the DVC model has a theoretically more powerful modeling capability through a richer form of basis functions. A novel basis function multiplexing-based model search algorithm, DVC search (DVCS), consisting of two main operations: nonlinear extension and phase rotation, is also proposed in this paper. The DVCS not only solves the practical deployment problem of the DVC model but also balances the performance and hardware complexity of the model. Two different power amplifiers (PAs) are used for comparison experiments to verify the performance of the DVCS model. The results of the experimental tests have demonstrated that the DVCS model can achieve an excellent trade-off between linearization performance and hardware complexity.

关键词： Behavioral modeling digital predistortion (DPD) greedy algorithm model complexity piecewise function power amplifiers (PAs) vector decomposition Volterra variant

来源：评论

学校读者我要写书评

暂无评论

Finding the appropriate level of complexity for a simulation model: An example with a forest growth model

引用

FOREST ECOLOGY AND MANAGEMENT 2008年第10期256卷 1659-1665页

作者： Astrup, Rasmus Coates, K. David Hall, Erin Bulkley Valley Ctr Nat Resources Res & Management Smithers BC V0J 2N0 Canada British Columbia Forest Serv Res Sect Smithers BC V0J 2N0 Canada

The topic of model complexity is fundamental to model developers and model users. In this study, we investigate how over- and under-fitting of a driving function in a simulation model influences the predictive ability of the model. Secondly, we investigate whether model selection approaches succeed in selecting driving functions with the best predictive ability. We address these issues through an example with the forest simulator SORTIE-ND. Utilizing maximum likelihood methods and individual tree growth data we parameterize five growth functions of increasing complexity. We then incorporate each growth function into the simulation model SORTIE-ND and test predicted growth against independent data. Compared to the independent data, the simplest and the most complex growth functions had the poorest predictive ability while functions of intermediate complexity had the best predictive ability. The poor predictive ability of the simplest model is caused by poor approximation of the system while the poor predictive ability of the most complex model is caused by biased parameter estimates. A growth function of intermediate complexity was the most parsimonious model where error due to approximation and error due to estimation were simultaneously minimized. The model selection criteria AIC and BIC were found to select complex functions that were over-fitted according to the independent data comparison. BIC was closer to choosing the model that minimized prediction error than AIC. In this example, BIC is the more appropriate model selection criterion. It is important that both model developers and models users remember that more complex models do not always result in better predictive models. (C) 2008 Elsevier B.V. All rights reserved.

关键词： Parsimony model complexity Growth Prediction SORTIE AIC BIC model selection

来源：评论

学校读者我要写书评

暂无评论

Gradient Information and Regularization for Gene Expression Programming to Develop Data-Driven Physics Closure models

引用

FLOW TURBULENCE AND COMBUSTION 2025年第1期114卷 145-175页

作者： Waschkowski, Fabian Li, Haochen Deshmukh, Abhishek Grenga, Temistocle Zhao, Yaomin Pitsch, Heinz Klewicki, Joseph Sandberg, Richard D. Univ Melbourne Dept Mech Engn Melbourne Vic 3010 Australia Peking Univ Ctr Appl Phys & Technol Beijing 100871 Peoples R China Rhein Westfal TH Aachen Inst Combust Technol D-52062 Aachen Germany

Learning accurate numerical constants when developing algebraic models is a known challenge for evolutionary algorithms, such as Gene Expression Programming (GEP). This paper introduces the concept of adaptive symbols to the GEP framework by Weatheritt and Sandberg (J Comput Phys 325:22-37, 2016a) to develop advanced physics closure models. Adaptive symbols utilize gradient information to learn locally optimal numerical constants during model training, for which we investigate two types of nonlinear optimization algorithms. The second contribution of this work is implementing two regularization techniques to incentivize the development of implementable and interpretable closure models. We apply L2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L_2$$\end{document} regularization to ensure small magnitude numerical constants and devise a novel complexity metric that supports the development of low complexity models via custom symbol complexities and multi-objective optimization. This extended framework is employed to four use cases, namely rediscovering Sutherland's viscosity law, developing laminar flame speed combustion models and training two types of fluid dynamics turbulence models. The model prediction accuracy and the convergence speed of training are improved significantly across all of the more and less complex use cases, respectively. The two regularization methods are essential for developing implementable closure models and we demonstrate that the developed turbulence models substantially improve simulations over state-of-the-art models.

关键词： Gene Expression Programming Nonlinear optimization Regularization model complexity

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：