Filters








3,681 Hits in 4.2 sec

Robust and Efficient Kernel Hyperparameter Paths with Guarantees

Joachim Giesen, Sören Laue, Patrick Wieschollek
2014 International Conference on Machine Learning  
Here we address this problem by devising a robust and efficient path tracking algorithm that can also handle kernel hyperparameter paths. The algorithm has asymptotically optimal complexity.  ...  kernel hyperparameter path.  ...  The kernel hyperparameter path for the IONOSPHERE data set with fixed bias updates (top, left) and dynamic bias updates (top, right), and the kernel hyperparameter path for the A1A data set with fixed  ... 
dblp:conf/icml/GiesenLW14 fatcat:3we3mx6ll5g5pblwwpgllr4x7m

Robust Policy Search for Robot Navigation with Stochastic Meta-Policies [article]

Javier Garcia-Barcos, Ruben Martinez-Cantin
2020 arXiv   pre-print
Then, to deal with mismodeling errors and improve exploration we use stochastic meta-policies for query selection and an adaptive kernel.  ...  We compare the proposed algorithm with previous results in several optimization benchmarks and robot tasks, such as pushing objects with a robot arm, or path finding with a rover.  ...  For the remainder of the paper we consider a GP with zero mean and kernel k : X × X → R with hyperparameters θ.  ... 
arXiv:2003.01000v1 fatcat:o7gs2bdenfaxlpja4rulbtuhee

Uncertainty-driven view planning for underwater inspection

Geoffrey A. Hollinger, Brendan Englot, Franz Hover, Urbashi Mitra, Gaurav S. Sukhatme
2012 2012 IEEE International Conference on Robotics and Automation  
We discuss the problem of inspecting an underwater structure, such as a submerged ship hull, with an autonomous underwater vehicle (AUV).  ...  Uncertainty modeling provides novel cost functions for planning the path of the AUV to minimize a metric of inspection performance.  ...  The authors gratefully acknowledge Jonathan Binney, Jnaneshwar Das, Arvind Pereira, and Hordur Heidarsson at the University of Southern California for their insightful comments.  ... 
doi:10.1109/icra.2012.6224726 dblp:conf/icra/HollingerEHMS12 fatcat:4fgujww4enarhmoprcqsgxa6qi

Safe Multi-Agent Interaction through Robust Control Barrier Functions with Learned Uncertainties [article]

Richard Cheng, Mohammad Javad Khojasteh, Aaron D. Ames, Joel W. Burdick
2020 arXiv   pre-print
Robots operating in real world settings must navigate and maintain safety while interacting with many heterogeneous agents and obstacles.  ...  We verify via simulation results that the nominal multi-agent CBF is often violated during agent interactions, whereas our robust formulation maintains safety with a much higher probability and adapts  ...  Fig. 2 : 2 Sample path of a multi-agent system based on the nominal CBF (cf. [6] ) and our proposed Robust CBF.  ... 
arXiv:2004.05273v2 fatcat:lqmwx54fd5hprkdihv5gf7s5qy

Funneled Bayesian Optimization for Design, Tuning and Control of Autonomous Systems [article]

Ruben Martinez-Cantin
2019 arXiv   pre-print
Bayesian optimization has become a fundamental global optimization algorithm in many problems where sample efficiency is of paramount importance.  ...  In order to generalize to unknown functions in a black-box fashion, the common assumption is that the underlying function can be modeled with a stationary process.  ...  ACKNOWLEDGMENT The authors would like to thank Javier García-Barcos for his help on the CFD simulator and Eduardo Montijano for his valuable comments.  ... 
arXiv:1610.00366v2 fatcat:3iwxcni4hvdgjl2pupx3jyjksm

Active planning for underwater inspection and the benefit of adaptivity

Geoffrey A Hollinger, Brendan Englot, Franz S Hover, Urbashi Mitra, Gaurav S Sukhatme
2012 The international journal of robotics research  
Such analysis allows the use of robust, non-adaptive planning algorithms that perform competitively with adaptive algorithms.  ...  We formulate the inspection planning problem as an extension to Bayesian active learning, and we show connections to recent theoretical guarantees in this area.  ...  Acknowledgments The authors gratefully acknowledge Jonathan Binney, Jnaneshwar Das, Arvind Pereira and Hordur Heidarsson at the University of Southern California for their insightful comments.  ... 
doi:10.1177/0278364912467485 fatcat:ksznirv22nd5dazct4blb7fzqu

Gaussian Process Uniform Error Bounds with Unknown Hyperparameters for Safety-Critical Applications [article]

Alexandre Capone, Armin Lederer, Sandra Hirche
2022 arXiv   pre-print
To mitigate this, we introduce robust Gaussian process uniform error bounds in settings with unknown hyperparameters.  ...  However, state-of-the-art techniques for safety-critical settings hinge on the assumption that the kernel hyperparameters are known, which does not apply in general.  ...  We thank Christian Fiedler for the useful comments and constructive feedback on the manuscript.  ... 
arXiv:2109.02606v2 fatcat:zknjqzsnsvfavlcbfkg4nthw5a

A Gaussian sampling heuristic estimation model for developing synthetic trip sets

S. F. A. Batista, Guido Cantelmo, Mónica Menéndez, Constantinos Antoniou
2021 Computer-Aided Civil and Infrastructure Engineering  
We show that the presented model is more robust and computationally efficient than the benchmark method.  ...  We also discuss how the choice of the kernel function and calibration of the hyperparameters influence the performance of the presented heuristic model.  ...  Second, we analyze the computational efficiency and robustness of the developed heuristic model in comparison to the benchmark method to determine synthetic sets of trips.  ... 
doi:10.1111/mice.12697 fatcat:xvnjlnnwtraa3jv3id4x3uho5m

Uncertainty-based Human Motion Tracking with Stable Gaussian Process State Space Models

Lukas Pöhler, Jonas Umlauft, Sandra Hirche
2019 IFAC-PapersOnLine  
We exploit the model fidelity which is related to the location of the training and test data: Our approach actively strives into regions with more demonstration data and thus higher model certainty.  ...  We exploit the model fidelity which is related to the location of the training and test data: Our approach actively strives into regions with more demonstration data and thus higher model certainty.  ...  The hyperparameters of the SE kernel are the lengthscales l j ∈ R + , j = 1, . . . , n.  ... 
doi:10.1016/j.ifacol.2019.01.002 fatcat:wmtduk2pqndble2h4njiub5nle

Prediction of Atomization Energy Using Graph Kernel and Active Learning [article]

Yu-Hang Tang, Wibe A. de Jong
2018 arXiv   pre-print
We then derive formulas for the efficient evaluation of the kernel.  ...  In this paper, we present a kernel-based pipeline that can learn and predict the atomization energy of molecules with high accuracy.  ...  Gaussian process regression (GPR) [6] is a robust regression method that features high accuracy, strong smoothness guarantee, and built-in uncertainty estimation.  ... 
arXiv:1810.07310v2 fatcat:b2sxpxaryjbjthbcgninjptn7i

Using Proximity Graph Cut for Fast and Robust Instance-Based Classification in Large Datasets

Stanislav Protasov, Adil Mehmood Khan
2021 Complexity  
This work presents the design and implementation of a classification algorithm with index data structures, which would allow us to build fast and scalable solutions for large multidimensional datasets.  ...  Our results suggest that the algorithm can be used in large-scale applications for fast and robust classification, especially when the search index is already constructed for the data.  ...  For implementation, we use both properties of NSW graphs to efficiently obtain a path in a graph and combine them with the Jordan curve theorem.  ... 
doi:10.1155/2021/2011738 doaj:2b877f8528d84394ba3f6fd5d2776363 fatcat:3pfkluar6zdvto5gsoehfwqnpq

A Neurochaos Learning Architecture for Genome Classification [article]

Harikrishnan NB and Pranay SY and Nithin Nagaraj
2020 arXiv   pre-print
These ChaosFEX features are then fed to a Support Vector Machine with linear kernel for classification.  ...  Robustness of ChaosFEX features to additive noise is also demonstrated.  ...  B. thanks "The University of Trans-Disciplinary Health Sciences and Technology (TDU)" for permitting this research as part of the PhD programme.  ... 
arXiv:2010.10995v1 fatcat:7dxjkbeoyfbf7ldd643fyebhqe

On-the-fly construction of surrogate constitutive models for concurrent multiscale mechanical analysis through probabilistic machine learning

I.B.C.M. Rocha, P. Kerfriden, F.P. van der Meer
2020 Journal of Computational Physics: X  
with robust probabilistic foundations.  ...  Hyperparameter optimization The process variance σ 2 f and length scale that compose the kernel and the target noise σ 2 n are hyperparameters that should be learned from the dataset D.  ... 
doi:10.1016/j.jcpx.2020.100083 fatcat:3ap2ocyz5ncv7gpz77rykqnpam

Multi-tasking the growth of cosmological structures [article]

Louis Perenon, Matteo Martinelli, Stéphane Ilić, Roy Maartens, Michelle Lochner, Chris Clarkson
2021 arXiv   pre-print
We find that this multi-task approach outperforms the single-task approach for future surveys and will allow us to detect departures from the standard model with higher significance.  ...  By contrast, the limited sensitivity of current data severely hinders the use of agnostic methods, since the Gaussian processes parameters need to be fine tuned in order to obtain robust reconstructions  ...  Cosmology with Radio Telescopes, ASTRO-0945), for providing computational resources for this research project.  ... 
arXiv:2105.01613v2 fatcat:57jztixyajhnfftxdofomxb3ke

Green Machine Learning via Augmented Gaussian Processes and Multi-Information Source Optimization [article]

Antonio Candelieri, Riccardo Perego, Francesco Archetti
2020 arXiv   pre-print
A comparison with a traditional Bayesian Optimization approach to optimize the hyperparameters of the SVM classifier on the large dataset only is reported.  ...  A strategy which has been gaining recently importance to drastically reduce computational time and energy consumed is to exploit the availability of different information sources, with different computational  ...  Compliance with Ethical Standards Ethical approval: This article does not contain any studies with human participants or animals performed by any of the authors.  ... 
arXiv:2006.14233v1 fatcat:5ivwye2n3nftdp44qj2kgzpila
« Previous Showing results 1 — 15 out of 3,681 results