Filters








182,907 Hits in 7.8 sec

Learning Multiple Models of Non-linear Dynamics for Control Under Varying Contexts [chapter]

Georgios Petkos, Marc Toussaint, Sethu Vijayakumar
2006 Lecture Notes in Computer Science  
This paper presents an efficient multiple model approach for non-linear dynamics, which can bootstrap context separation from context-unlabeled data and realizes simultaneous online context estimation,  ...  The approach formulates a consistent probabilistic model used to infer the unobserved context and uses Locally Weighted Projection Regression as an efficient online regressor which provides local confidence  ...  In this paper we present a non-linear multiple model approach to control based on an efficient non-linear online learning algorithm (LWPR) that addresses these requirements.  ... 
doi:10.1007/11840817_93 fatcat:xxknbi4q6reopnxg6zbz4db6ay

Learning symmetric face pose models online using locally weighted projectron regression

Jawad Nagi, Gianni A. Di Caro, Alessandro Giusti, Luca M. Gambardella
2014 2014 IEEE International Conference on Image Processing (ICIP)  
Using the Locally Weighted Projectron Regression (LWPR), an online incremental regression-based learning scheme, we can reliably learn and predict the pose of a human face in real-time at a low computational  ...  This paper proposes a simple and efficient approach for estimating the distance and orientation of an human, from a single robot-acquired image.  ...  For online learning and prediction of the face pose, we augment the face quality measures with the Locally Weighted Projectron Regression (LWPR), an online regression approach, that uses a mixture of locally  ... 
doi:10.1109/icip.2014.7025280 dblp:conf/icip/NagiCGG14 fatcat:fteof6744vbqrfx3ktpblbtci4

Collaborative Unsupervised Visual Representation Learning from Decentralized Data [article]

Weiming Zhuang, Xin Gan, Yonggang Wen, Shuai Zhang, Shuai Yi
2021 arXiv   pre-print
As such, a natural problem is how to leverage these data to learn visual representations for downstream tasks while preserving data privacy.  ...  In this framework, each party trains models from unlabeled data independently using contrastive learning with an online network and a target network.  ...  The state-of-the-art unsupervised learning approaches are effective, but they may not work well with non-IID data, as shown in Figure 1 (a).  ... 
arXiv:2108.06492v1 fatcat:hkunizhgongzzgg26xu55kgesi

Is Non-IID Data a Threat in Federated Online Learning to Rank? [article]

Shuyi Wang, Guido Zuccon
2022 arXiv   pre-print
In this perspective paper we study the effect of non independent and identically distributed (non-IID) data on federated online learning to rank (FOLTR) and chart directions for future work in this new  ...  While FOLTR systems are on their own rights a type of federated learning system, the presence and effect of non-IID data in FOLTR has not been studied.  ...  ACKNOWLEDGMENTS We would like to thank the anonymous reviewers for their insightful feedback in further shaping the paper.  ... 
arXiv:2204.09272v2 fatcat:ea5yvasuf5clheq7wdjznzexye

Real-time learning of resolved velocity control on a Mitsubishi PA-10

Jan Peters, Duy Nguyen-Tuong
2008 2008 IEEE International Conference on Robotics and Automation  
While this problem can be treated in various ways in offline learning, it poses a serious problem for online learning.  ...  While humans acquire this transformation to complicated tool spaces with ease, it is not a straightforward application for supervised learning algorithms due to non-convex learning problem.  ...  For doing so, we had to overcome the difficulties of having a non-convex data distribution by only learning in the vicinity of a local model anchored both in joint position.  ... 
doi:10.1109/robot.2008.4543645 dblp:conf/icra/PetersN08 fatcat:i6oyuzxctvgdhnwtq7ivbyhdym

Online Open World Recognition [article]

Rocco De Rosa, Thomas Mensink, Barbara Caputo
2016 arXiv   pre-print
Experimentally we validate our approach on two large-scale datasets in different learning scenarios. For all these scenarios our proposed methods outperform their non-online counterparts.  ...  We conclude that local and online learning is important to capture the full dynamics of open world recognition.  ...  In the next section we introduce a local learning approach which allows for non-linear classification.  ... 
arXiv:1604.02275v1 fatcat:nepamsrsjzcinfttfyguzyb4ia

Online Learning for Hierarchical Networks of Locally Arranged Models using a Support Vector Domain Model

Florian Hoppe, Gerald Sommer
2007 Neural Networks (IJCNN), International Joint Conference on  
Secondly, an online learning algorithm for our approach will be described that can be used in applications where training data is only available as a continuous stream of samples.  ...  We propose two new developments for our supervised local linear approximation technique, the so called Hierarchical Network of Locally Arranged Models.  ...  However, this paper does not necessarily represent the opinion of the European Community, and the European Community is not responsible for any use which may be made of its contents.  ... 
doi:10.1109/ijcnn.2007.4370966 dblp:conf/ijcnn/HoppeS07 fatcat:tijrcdxyv5ayrhzvzvu4ox55fi

Data-driven multi-model control for a waste heat recovery system

Johan Peralez, Francesco Galuppo, Pascal Dufour, Christian Wolf, Madiha Nadri
2020 2020 59th IEEE Conference on Decision and Control (CDC)  
We consider the problem of supervised learning of a multi-model based controller for non-linear systems.  ...  Selected multiple linear controllers are used for different operating points and combined with a local weighting scheme, whose weights are predicted by a deep neural network trained online.  ...  Approaches based on Non-Linear models: global results, hard to design.  ... 
doi:10.1109/cdc42340.2020.9304418 fatcat:olotltj76fgnfdzkteoyq5h3mq

FedSiam: Towards Adaptive Federated Semi-Supervised Learning [article]

Zewei Long, Liwei Che, Yaqing Wang, Muchao Ye, Junyu Luo, Jinze Wu, Houping Xiao, Fenglong Ma
2021 arXiv   pre-print
FedSiam is built upon a siamese network into FL with a momentum update to handle the non-IID challenges introduced by unlabeled data.  ...  We further propose a new metric to measure the divergence of local model layers within the siamese network.  ...  Thus, how to utilize unlabeled data residing on local clients to learn the global model is a new challenge for FL.  ... 
arXiv:2012.03292v2 fatcat:bmurr4mrpzbrrkpqww6jtpzs6a

Learning task-space tracking control with kernels

Duy Nguyen-Tuong, Jan Peters
2011 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems  
In this paper, we use this insight to formulate a local kernelbased learning approach for online model learning for taskspace tracking control.  ...  Here, data driven learning methods may present an interesting alternative approach. However, learning models for task-space tracking control from sampled data is an ill-posed problem.  ...  First, we show for a toy example how a non-unique function can be learned in the online setting using this local learning approach.  ... 
doi:10.1109/iros.2011.6048038 fatcat:jfn5sqvgmfbtvill4745yo4voe

Learning task-space tracking control with kernels

Duy Nguyen-Tuong, J. Peters
2011 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems  
In this paper, we use this insight to formulate a local kernelbased learning approach for online model learning for taskspace tracking control.  ...  Here, data driven learning methods may present an interesting alternative approach. However, learning models for task-space tracking control from sampled data is an ill-posed problem.  ...  First, we show for a toy example how a non-unique function can be learned in the online setting using this local learning approach.  ... 
doi:10.1109/iros.2011.6094428 dblp:conf/iros/Nguyen-TuongP11 fatcat:mxpae4h7bfbqpdjc5rqgnx32ge

Online Kernel-Based Learning for Task-Space Tracking Robot Control

Duy Nguyen-Tuong, J. Peters
2012 IEEE Transactions on Neural Networks and Learning Systems  
In this paper, we use this insight to formulate a local, kernel-based learning approach for online model learning for task-space tracking control.  ...  Here, data driven model learning methods may present an interesting alternative approach. However, learning models for task-space tracking control from sampled data is an illposed problem.  ...  First, we show for a toy example how a non-unique function can be learned in the online setting using this local learning approach.  ... 
doi:10.1109/tnnls.2012.2201261 pmid:24807925 fatcat:w5bnhjo57fh67euhegyulystjq

Efficient online learning of a non-negative sparse autoencoder

Andre Lemme, René Felix Reinhart, Jochen Jakob Steil
2010 The European Symposium on Artificial Neural Networks  
We introduce an efficient online learning mechanism for nonnegative sparse coding in autoencoder neural networks.  ...  In this paper we compare the novel method to the batch algorithm non-negative matrix factorization with and without sparseness constraint.  ...  Both online learning rules are combined, use only local information, and are very efficient.  ... 
dblp:conf/esann/LemmeRS10 fatcat:ljghbae2crcn5jrycd5fkxoibq

Real-Time Online Adaptive Feedforward Velocity Control for Unmanned Ground Vehicles [chapter]

Nicolai Ommer, Alexander Stumpf, Oskar von Stryk
2018 Lecture Notes in Computer Science  
This paper proposes a new adaptive compensation feedforward controller capable of online learning a compensation motion model without any prior knowledge to counteract non-modeled disturbance such as slippage  ...  The controller is able to prevent motion errors a priori and is well suited for real hardware due to high adaptation rate.  ...  Locally Weighted Learning (LWL) categorizes a class of non-linear function learners based on locally linear approximations.  ... 
doi:10.1007/978-3-030-00308-1_1 fatcat:uoe6jatnavfefgd67yowqiauge

Learning tracking control with forward models

Botond Bocsi, Philipp Hennig, Lehel Csato, Jan Peters
2012 2012 IEEE International Conference on Robotics and Automation  
We propose an adaptive learning algorithm for tracking control of underactuated or non-rigid robots where the physical model of the robot is unavailable.  ...  The control method is based on the fact that forward models are relatively straightforward to learn and local inversions can be obtained via local optimization.  ...  Learning Inverse Kinematics Among the most well-known online learning approaches for inverse kinematics is D'Souza et. al. [4] , based on "locally weighted projection regression" (LWPR) [21] .  ... 
doi:10.1109/icra.2012.6224831 dblp:conf/icra/BocsiHCP12 fatcat:wuzem7owwfdgja7xmtjpadiy4a
« Previous Showing results 1 — 15 out of 182,907 results