A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf
.
Filters
Network Independent Rates in Distributed Learning
[article]
2015
arXiv
pre-print
Network independent rates were not available for other consensus based distributed learning algorithms. ...
Our main result states that, after a transient time, all agents will concentrate their beliefs at a network independent rate. ...
network independent convergence rate. ...
arXiv:1509.08574v1
fatcat:xaiw4qc3rzfgrfuayhaw2idsuu
An Algorithm to Learn Causal Relations Between Genes from Steady State Data: Simulation and Its Application to Melanoma Dataset
[chapter]
2005
Lecture Notes in Computer Science
regulatory network, for jointly learning the causal relationship among genes. ...
In this paper, we present a modified IC (mIC) algorithm that uses entropy to test conditional independence and combines the steady state data with partial prior knowledge of topological ordering in gene ...
G and a distribution P are faithful to each other if they exhibit the same set of independencies. ...
doi:10.1007/11527770_69
fatcat:34sf42ialzfrraltovz4l6jld4
Fast Feature Fool: A data independent approach to universal adversarial perturbations
[article]
2017
arXiv
pre-print
In this paper, for the first time, we propose a novel data independent approach to generate image agnostic perturbations for a range of CNNs trained for object recognition. ...
In the absence of data, our method generates universal adversarial perturbations efficiently via fooling the features learned at multiple layers thereby causing CNNs to misclassify. ...
Let X denote the distribution of images in R d and f denotes the classification function learned by a CNN that maps an image x ∼ X from the distribution to an estimated label f (x). ...
arXiv:1707.05572v1
fatcat:gdetsfxpvvak5nykqtdiwfysfy
The BCM rule allows a spinal cord model to learn rhythmic movements
[article]
2021
bioRxiv
pre-print
to explain learning in the visual cortex. ...
Here we demonstrate that rhythmic and alternating movements in pendulum models can be learned by a monolayer spinal cord circuitry model using the BCM learning rule, which has been previously proposed ...
In addition, after the learning the voltage distributions of three of the eight neurons for the independent pendulums system were clearly bimodal, in line with the expected result of the BCM learning rule ...
doi:10.1101/2021.11.12.467473
fatcat:6fxnq4wluvhabk3m5h4vjr77fe
Asymptotic Network Independence in Distributed Stochastic Optimization for Machine Learning
[article]
2020
arXiv
pre-print
We provide a discussion of several recent results which, in certain scenarios, are able to overcome a barrier in distributed stochastic optimization for machine learning. ...
Our focus is the so-called asymptotic network independence property, which is achieved whenever a distributed method executed over a network of n nodes asymptotically converges to the optimal solution ...
Acknowledgments We would like to thank Artin Spiridonoff from Boston University for his kind help in providing Figure 3 . ...
arXiv:1906.12345v5
fatcat:ujwkgyfcezffrjjwhit5lldqim
Blended Learning Innovation Model among College Students Based on Internet
2018
International Journal of Emerging Technologies in Learning (iJET)
It can be discovered from this research that the blended learning model is superior to the single and traditional teaching mode or the pure network teaching mode in the aspects of in-spiring the learning ...
in the blended learning is low. ...
indicates that the college students are weak in managing the study progress in network independent learning with an average score of 2.9189, as shown by the specific distribution in Figure 8 . ...
doi:10.3991/ijet.v13i10.9454
fatcat:iiqflwncrbg7jfj5ogbihrhf6u
Eigenspace Method by Autoassociative Networks for Object Recognition
[chapter]
2004
Lecture Notes in Computer Science
Five layered autoassociative network is available to obtain a manifold on the minimum square error hypersurface which approximates a distribution of learning sample. ...
networks. ...
In contrast with the pattern recognition using the mutual associative networks, each autoassociative network is organized independently for each class, and the training load of the networks can be distributed ...
doi:10.1007/978-3-540-27868-9_9
fatcat:7pwmyygvrvbphem25ma37sj4bq
Cooperation of spike timing-dependent and heterosynaptic plasticities in neural networks: A Fokker-Planck approach
2006
Chaos
It is believed that both Hebbian and homeostatic mechanisms are essential in neural learning. ...
Based on the Fokker-Planck theory and extensive numerical computations, we demonstrate that HSP and STDP operated on different time scales can complement each other, resulting in more realistic network ...
Figure 6͑a͒ shows the firing rate distribution of the excitatory network neurons with independent inputs, or C =0. ...
doi:10.1063/1.2189969
pmid:16822008
fatcat:bna2pumwdzcz7eyu5tynhmd3um
Evolving neural networks: Is it really worth the effort?
2005
The European Symposium on Artificial Neural Networks
The idea of using simulated evolution to create neural networks that learn faster and generalize better is becoming increasingly widespread. ...
from existing evolved networks that can be applied directly to our hand-crafted networks. ...
Conclusions Evolution has clearly shown that having four independent learning rates and initial weight distributions results in far superior learning performance compared with just one for the whole network ...
dblp:conf/esann/Bullinaria05
fatcat:enacm3iv2vb7zn27ohhhlbf3rq
Hyperplane Arrangements of Trained ConvNets Are Biased
[article]
2020
arXiv
pre-print
We investigate the geometric properties of the functions learned by trained ConvNets in the preactivation space of their convolutional layers, by performing an empirical study of hyperplane arrangements ...
We introduce statistics over the weights of a trained network to study local arrangements and relate them to the training dynamics. ...
learning rate. ...
arXiv:2003.07797v1
fatcat:dwarqycnynfb7onraczy6rphee
Surges of Collective Human Activity Emerge from Simple Pairwise Correlations
2019
Physical Review X
In addition to providing accurate quantitative predictions, we show that the topology of learned Ising interactions resembles the network of inter-human communication within a population. ...
Human populations exhibit complex behaviors---characterized by long-range correlations and surges in activity---across a range of social, political, and technological contexts. ...
Surges of human activity and failure of the independent approximation. (a) Distribution of interevent times for individuals in a network of email correspondence. ...
doi:10.1103/physrevx.9.011022
fatcat:cfya3nyr7zconfnt5qgekr7ksu
Risk-utility tradeoff shapes memory strategies for evolving patterns
[article]
2021
arXiv
pre-print
However, learning and memory storage for dynamic patterns still pose challenges in machine learning. Here, we introduce an analytical energy-based framework to address this problem. ...
Our approach offers a general guideline for learning and memory storage in systems interacting with diverse and evolving stimuli. ...
Irrespective of the network structure, an increase in learning rate is necessary for a network to follow, recognize, and store effective memory of evolving patterns [9] . ...
arXiv:2110.15008v1
fatcat:ozyk64qpyneanfd654coxwzgyq
Risk-utility tradeoff shapes memory strategies for evolving patterns
[article]
2021
bioRxiv
pre-print
However, learning and memory storage for dynamic patterns still pose challenges in machine learning. Here, we introduce an analytical energy-based framework to address this problem. ...
Our approach offers a general guideline for learning and memory storage in systems interacting with diverse and evolving signals. ...
Irrespective of the network structure, an increase in learning rate is necessary for a network to follow, recognize, and store effective memory of evolving patterns [9] . ...
doi:10.1101/2021.10.27.466120
fatcat:sordqbce5vgf5f2oym67n5d33m
Learning sensory representations with intrinsic plasticity
2007
Neurocomputing
In this paper we show how a network of such units can solve a standard non-linear independent component analysis (ICA) problem. ...
sensory representations in the cortex. r ...
Even in conditions No IP and Linear, Gabor-like receptive fields will develop in the network, but at a dramatically slower rate. ...
doi:10.1016/j.neucom.2006.11.006
fatcat:egbionfjjrdxrk4vcpdxirkoda
Metropolis-Hastings view on variational inference and adversarial training
[article]
2019
arXiv
pre-print
To address this question, we propose to learn an independent sampler that maximizes the acceptance rate of the MH algorithm, which, as we demonstrate, is highly related to the conventional variational ...
In particular, we demonstrate improvements of recently proposed BigGAN model on ImageNet. ...
In case of independent proposal distribution we show that the acceptance rate defines a semimetric in distribution space between p and q (see Appendix B). ...
arXiv:1810.07151v2
fatcat:526fulgwgnhefmhkjbvlaxdwea
« Previous
Showing results 1 — 15 out of 748,845 results