检索结果-内蒙古大学图书馆

Proceedings of the 34th International Conference on Neural Information Processing Systems

作者： Akshunna S. Dogra William T. Redman John A. Paulson School of Engineering and Applied Sciences Harvard University Cambridge MA Interdeparmental Graduate Program in Dynamical Neuroscience University of California Santa Barbara Santa Barbara CA

ISBN: (纸本)9781713829546

Koopman operator theory, a powerful framework for discovering the underlying dynamics of nonlinear dynamical systems, was recently shown to be intimately connected with neural network training. In this work, we take the first steps in making use of this connection. As Koopman operator theory is a linear theory, a successful implementation of it in evolving network weights and biases offers the promise of accelerated training, especially in the context of deep networks, where optimization is inherently a non-convex problem. We show that Koopman operator theoretic methods allow for accurate predictions of weights and biases of feedforward, fully connected deep networks over a non-trivial range of training time. During this window, we find that our approach is >10x faster than various gradient descent based methods (e.g. Adam, Adadelta, Adagrad), in line with our complexity analysis. We end by highlighting open questions in this exciting intersection between dynamical systems and neural network theory, and additional methods by which our results may be generalized.

关键词：

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

Local error quantification for efficient neural network dynamical system solvers

arXiv

引用

arXiv 2020年

作者： Dogra, Akshunna S. Redman, William T. John A. Paulson School of Engineering and Applied Sciences Harvard University CambridgeMA02138 United States Interdepartmental Graduate Program in Dynamical Neuroscience University of California Santa BarbaraCA93106 United States

Neural Networks have been identified as potentially powerful tools for the study of complex systems. A noteworthy example is the Neural Network Differential Equation (NN DE) solver, which can provide functional approximations to the solutions of a wide variety of differential equations. However, there is a lack of work on the role precise error quantification can play in their predictions: most variants focus on ambiguous and/or global measures of performance like the loss function. We address this in the context of dynamical system NN DE solvers, leveraging their learnt information to develop more accurate and efficient solvers, while still pursuing a completely unsupervised, data free approach. We achieve this by providing methods for quantifying the performance of NN DE solvers at local scales, thus allowing the user the capacity for efficient and targeted error correction. We showcase the utility of these methods by testing them on a nonlinear and a chaotic system each. Copyright © 2020, The Authors. All rights reserved.

关键词： Chaotic systems

来源：评论

学校读者我要写书评

暂无评论

Optimizing neural networks via Koopman operator theory

arXiv

引用

arXiv 2020年

作者： Dogra, Akshunna S. Redman, William T. John A. Paulson School of Engineering and Applied Sciences Harvard University CambridgeMA02138 United States Interdeparmental Graduate Program in Dynamical Neuroscience University of California Santa Barbara Santa BarbaraCA93106 United States

Koopman operator theory, a powerful framework for discovering the underlying dynamics of nonlinear dynamical systems, was recently shown to be intimately connected with neural network training. In this work, we take the first steps in making use of this connection. As Koopman operator theory is a linear theory, a successful implementation of it in evolving network weights and biases offers the promise of accelerated training, especially in the context of deep networks, where optimization is inherently a non-convex problem. We show that Koopman operator theoretic methods allow for accurate predictions of weights and biases of feedforward, fully connected deep networks over a non-trivial range of training time. During this window, we find that our approach is >10x faster than various gradient descent based methods (e.g. Adam, Adadelta, Adagrad), in line with our complexity analysis. We end by highlighting open questions in this exciting intersection between dynamical systems and neural network theory. We highlight additional methods by which our results could be expanded to broader classes of networks and larger training intervals, which shall be the focus of future work. Copyright © 2020, The Authors. All rights reserved.

关键词： Nonlinear dynamical systems

来源：评论

学校读者我要写书评

暂无评论

Control of ecological outcomes through deliberate parameter changes in a model of the gut microbiome

arXiv

引用

arXiv 2019年

作者： Wang, Zipeng Jones, Eric W. Mueller, Joshua M. Carlson, Jean M. Department of Physics University of California Santa Barbara Santa BarbaraCA United States Interdepartmental Graduate Program in Dynamical Neuroscience University of California Santa Barbara Santa BarbaraCA United States

The generalized Lotka-Volterra (gLV) equations are a mathematical proxy for ecological dynamics. We focus on a gLV model of the gut microbiome, in which the evolution of the gut microbial state is determined in part by pairwise inter-species interaction parameters that encode environmentally-mediated resource competition between microbes. We develop an in silico method that controls the steady-state outcome of the system by adjusting these interaction parameters. This approach is confined to a bistable region of the gLV model. The two steady states of interest are idealized as either a"healthy" or "diseased" steady state of the gut microbiome. In this method, a dimensionality reduction technique called steady-state reduction (SSR) is first used to generate a two-dimensional (2D) gLV model that approximates the high-dimensional dynamics on the 2D subspace spanned by the two steady states. Then a bifurcation analysis of the 2D model analytically determines parameter modifications that drive a disease-prone initial condition to the healthy steady state. This parameter modification of the reduced 2D model guides parameter modifications of the original high-dimensional model, resulting in a change of steady-state outcome in the high-dimensional model. This control method, called SPARC (SSR-guided parameter change), bypasses the computational challenge of directly determining parameter modifications in the original high-dimensional system. SPARC could guide the development of indirect bacteriotherapies, which seek to change microbial compositions by deliberately modifying gut environmental variables such as gut acidity or macronutrient availability. Copyright © 2019, The Authors. All rights reserved.

关键词： Bifurcation (mathematics)

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：