Filters








1,550,191 Hits in 6.7 sec

Decoupling Gradient-Like Learning Rules from Representations

Philip S. Thomas, Christoph Dann, Emma Brunskill
2018 International Conference on Machine Learning  
When creating a machine learning system, we must make two decisions: what representation should be used (i.e., what parameterized function should be used) and what learning rule should be used to search  ...  That is, using the same learning rule with two different representations that can represent the same sets of functions can result in two different outcomes.  ...  ) , noticed that a natural gradient method (using the Fisher information matrix for G(θ), which results in a first-order covariant update with respect to the set of parameterized probability distributions  ... 
dblp:conf/icml/ThomasDB18 fatcat:buzow7df6jbdnfb4qtjnvgulmu

Optimization for Data-Driven Learning and Control

Usman A. Khan, Waheed U. Bajwa, Angelia Nedic, Michael G. Rabbat, Ali H. Sayed
2020 Proceedings of the IEEE  
Decentralized and distributed first-order methods are discussed, which have found success in many emerging machine learning problems where they serve as the first method of choice for many complex inference  ...  A General Framework for Decentralized Optimization With First-Order Methods by R. Xin, S. Pu, A. Nedić, and U. A.  ... 
doi:10.1109/jproc.2020.3031225 fatcat:6ibimo2s2zgepbyeya2fjq7flu

ArchNet: Data Hiding Model in Distributed Machine Learning System [article]

Kaiyan Chang, Wei Jiang, Jinyu Zhan, Zicheng Gong, Weijia Pan
2020 arXiv   pre-print
It also shows that ArchNet can be deployed on the distributed system with embedded devices.  ...  In this paper, we approach to address the data hiding problem in such distributed machine learning systems.  ...  In order to solve this problem, we propose distributed machine learning system which is different from the distributed machine learning on heterogeneous computing system (i.e, a computer with GPU, TPU  ... 
arXiv:2004.10968v2 fatcat:ngitxy4rkzbkdl5onnfhv7j4la

Communication-Efficient Edge AI: Algorithms and Systems [article]

Yuanming Shi, Kai Yang, Tao Jiang, Jun Zhang, Khaled B. Letaief
2020 arXiv   pre-print
Specifically, we first identify key communication challenges in edge AI systems.  ...  This is driven by the explosive growth of data, advances in machine learning (especially deep learning), and easy access to vastly powerful computing resources.  ...  bandwidth as first-order distributed learning algorithms.  ... 
arXiv:2002.09668v1 fatcat:nhasdzb7t5dt5brs2r7ocdzrnm

Table of Contents

2020 Proceedings of the IEEE  
|INVITED PAPER| This article presents a collection of state-of-the-art results for distributed optimization problems arising in the context of robot networks, with a focus on two special classes of problems  ...  |INVITED PAPER| This article presents a general framework for distributed firstorder methods, for minimizing a finite sum of functions, that is applicable to both undirected and directed graphs.  ...  2067 Accelerated First-Order Optimization Algorithms for Machine Learning By H. Li, C. Fang, and Z. Lin  ... 
doi:10.1109/jproc.2020.3028590 fatcat:bwlj7gfvcrbnfgkxihjmn2dssa

Active Object Localization in Visual Situations [article]

Max H. Quinn, Anthony D. Rhodes, Melanie Mitchell
2016 arXiv   pre-print
More specifically, the system learns a set of probability distributions describing spatial and other relationships among relevant objects.  ...  Finally, we contrast our method with several other approaches that use context as well as active search for object localization in images.  ...  Many thanks to Li-Yun Wang for running the experiments with Randomized Prim's algorithm. This material is based upon work supported by the National Science Foundation under Grant Number IIS-1423651.  ... 
arXiv:1607.00548v1 fatcat:h3b54sulizdephind6upw6cfbi

A Robust Transfer Dictionary Learning Algorithm for Industrial Process Monitoring

Chunhua Yang, Huiping Liang, Keke Huang, Yonggang Li, Weihua Gui
2021 Engineering  
Data-driven process-monitoring methods have been the mainstream for complex industrial systems due to their universality and the reduced need for reaction mechanisms and first-principles knowledge.  ...  proposed to deal with the problem of the distribution divergence of realistic industrial process monitoring.  ...  Compliance with ethics guidelines Chunhua Yang, Huiping Liang, Keke Huang, Yonggang Li, and Weihua Gui declare that they have no conflict of interest or financial conflicts to disclose.  ... 
doi:10.1016/j.eng.2020.08.028 fatcat:wubwrribo5hkrgok4q653wvpry

A Survey on Neural-symbolic Systems [article]

Dongran Yu, Bo Yang, Dayou Liu, Hui Wang
2021 arXiv   pre-print
In this case, an ideal intelligent system--a neural-symbolic system--with high perceptual and cognitive intelligence through powerful learning and reasoning capabilities gains a growing interest in the  ...  This paper surveys the latest research in neural-symbolic systems along four dimensions: the necessity of combination, technical challenges, methods, and applications.  ...  The first category is based on the symbolic system with a supplementary neural system.  ... 
arXiv:2111.08164v1 fatcat:bc33afiitnb73bmjtrfbdgkwpy

Can machines solve general queueing systems? [article]

Eliran Sherzer, Arik Senderovich, Opher Baron, Dmitry Krass
2022 arXiv   pre-print
To the best of our knowledge, this is the first time a machine learning model is applied to a general queueing theory problem.  ...  This shows the promise of extending our approach beyond the analytically solvable systems (e.g., G/G/1 or G/G/c).  ...  For example in [3] , the authors present an explicit method to compose minimal order continuous-time acyclic phase type (APH) distributions given only the first three moments.  ... 
arXiv:2202.01729v1 fatcat:sqq5s3k4hjaqrhyhfgimz6xoii

Neural network enhanced hybrid quantum many-body dynamical distributions

Rouven Koch, Jose L. Lado
2021 Physical Review Research  
Focusing on many-body dynamical distributions, we show that this hybrid neural-network many-body algorithm, trained with single-particle data only, can efficiently extrapolate dynamics for many-body systems  ...  Computing dynamical distributions in quantum many-body systems represents one of the paradigmatic open problems in theoretical condensed matter physics.  ...  In order to train network, we are using the supervised learning based on backpropagation.  ... 
doi:10.1103/physrevresearch.3.033102 fatcat:crj2ezu7ijekpkrfxpzxcn3tju

Decoupling Learning Rules from Representations [article]

Philip S. Thomas and Christoph Dann and Emma Brunskill
2017 arXiv   pre-print
That is, using the same learning rule with two different representations that can represent the same sets of functions can result in two different outcomes.  ...  When creating an artificial intelligence system, we must make two decisions: what representation should be used (i.e., what parameterized function should be used) and what learning rule should be used  ...  The learning rules that our method can transform into first-order covariant learning rules satisfy: Assumption 1.  ... 
arXiv:1706.03100v1 fatcat:7lkegq3qqje7hoeihg3hsp62ia

Practical method for blind inversion of Wiener systems

2004 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat No 04CH37541) IJCNN-04  
Secondly, we review the Gaussianization method for blind inversion of Wiener systems.  ...  In this paper, firstly we show that the problem of blind inversion of Wiener systems is a special case of blind separation of post-nonlinear instantaneous mixtures approximately, and derive the learning  ...  Hence in order to analyze the performance of our method, we first investige the property of the C-F expansion.  ... 
doi:10.1109/ijcnn.2004.1380954 fatcat:nfzlinsf5zc7zhc4vjhaaezwfa

Reweighted Autoencoded Variational Bayes for Enhanced Sampling (RAVE) [article]

Joao Marcelo Lamim Ribeiro, Pablo Bravo Collado, Yihang Wang, Pratyush Tiwary
2018 arXiv   pre-print
RAVE involves iterations between molecular simulations and deep learning in order to produce an increasingly accurate probability distribution along a low-dimensional latent space that captures the key  ...  Here we propose the Reweighted Autoencoded Variational Bayes for Enhanced Sampling (RAVE) method, a new iterative scheme that uses the deep learning framework of variational autoencoders to enhance sampling  ...  We are hopeful this method will add a new tool in the exploration of complex molecular systems plagued with rare events. ACKNOWLEDGMENTS PT thanks Dr.  ... 
arXiv:1802.03420v1 fatcat:hgrdaacswfaprlax72dkjfvnci

Reweighted autoencoded variational Bayes for enhanced sampling (RAVE)

João Marcelo Lamim Ribeiro, Pablo Bravo, Yihang Wang, Pratyush Tiwary
2018 Journal of Chemical Physics  
RAVE involves iterations between molecular simulations and deep learning in order to produce an increasingly accurate probability distribution along a low-dimensional latent space that captures the key  ...  Here we propose the Reweighted Autoencoded Variational Bayes for Enhanced Sampling (RAVE) method, a new iterative scheme that uses the deep learning framework of variational autoencoders to enhance sampling  ...  We are hopeful this method will add a new tool in the exploration of complex molecular systems plagued with rare events. ACKNOWLEDGMENTS PT thanks Dr.  ... 
doi:10.1063/1.5025487 pmid:30134694 fatcat:2qxaxfmcwvfm5bnkbd3hypq5n4

Design Of An Intelligent Tutor Using A Multiagent Approach

Kamel Khoualdi, Radia Benghezal
2007 Zenodo  
Research in distributed artificial intelligence and multiagent systems consider how a set of distributed entities can interact and coordinate their actions in order to solve a given problem.  ...  The second one deals particularly with the design of a part of a tutor system: the pedagogue agent.  ...  We can for example classify these plans according to three criteria: --Method of a lesson presentation: a detailed method with examples or a simplified method with only definitions (detailed, simplified  ... 
doi:10.5281/zenodo.1080633 fatcat:va6awty6v5a5hoxbz54tzu67ei
« Previous Showing results 1 — 15 out of 1,550,191 results