Filters








77,958 Hits in 4.7 sec

Real-Time Local GP Model Learning [chapter]

Duy Nguyen-Tuong, Matthias Seeger, Jan Peters
2010 Studies in Computational Intelligence  
., standard GPR, support vector regression (SVR) and locally weighted projection regression (LWPR), show that LGP has high approximation accuracy while being sufficiently fast for real-time online learning  ...  In this paper, we propose an approximation to the standard GPR using local Gaussian processes models inspired by [1, 2] .  ...  It also interesting to investigate further applications of the LGP in humanoid robotics with 35 of more DoFs and learning other types of the control such as operational space control.  ... 
doi:10.1007/978-3-642-05181-4_9 fatcat:liqxstsdxbftzbqupzcyoanw2q

Real-Time Regression with Dividing Local Gaussian Processes [article]

Armin Lederer, Alejandro Jose Ordonez Conejo, Korbinian Maier, Wenxin Xiao, Jonas Umlauft, Sandra Hirche
2021 arXiv   pre-print
Therefore, this paper proposes dividing local Gaussian processes, which are a novel, computationally efficient modeling approach based on Gaussian process regression.  ...  shows various favorable theoretical properties (uncertainty estimate, unlimited expressive power), the poor scaling with respect to the training set size prohibits its application in big data regimes in real-time  ...  Dividing Local Gaussian Processes While existing local GP approaches for real-time learning base on the principle that all local models have the same spatial extension in the input domain, our proposed  ... 
arXiv:2006.09446v2 fatcat:un2kgmd3tzf7rgxmzpouf6jlya

Adaptive Sensing for Learning Nonstationary Environment Models [article]

Sahil Garg, Amarjeet Singh, Fabio Ramos
2018 arXiv   pre-print
The core idea in LISAL is to learn two models using Gaussian processes (GPs) wherein the first is a nonstationary GP directly modeling the phenomenon.  ...  The second model uses a stationary GP representing a latent space corresponding to changes in dynamics, or the nonstationarity characteristics of the first model.  ...  We first learn a stationary GP model, with hyperparameters θ z0 , using the real observations (Line 3 in Algorithm 1).  ... 
arXiv:1804.10279v1 fatcat:vhph4bhycbgvnjx6zhwhl2z33q

Approximate real-time optimal control based on sparse Gaussian process models

Joschka Boedecker, Jost Tobias Springenberg, Jan Wulfing, Martin Riedmiller
2014 2014 IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL)  
This results in an extremely dataefficient learning algorithm that can operate under real-time constraints.  ...  Our algorithm jointly learns a non-parametric model of the system dynamics -based on Gaussian Process Regression (GPR) -and performs receding horizon control using an adapted iterative LQR formulation.  ...  We thank Christof Schoetz for help with the implementation, Thomas Lampe for help with the real cart-pole system, Manuel Blum for developing libgp, and the three anonymous reviewers for helpful comments  ... 
doi:10.1109/adprl.2014.7010608 dblp:conf/adprl/BoedeckerSWR14 fatcat:6yegxwxt6nh3vctaqfiqsusj2q

Online Sparse Gaussian Process (GP) Regression Model For Human Motion Tracking

Sri Lavanya Sajja
2012 IOSR Journal of Engineering  
This model is a combination of temporal and spatial local GP experts model for efficient estimation of human pose.  ...  In this paper, we use a new technique known as online sparse Gaussian Process (GP) regression model.  ...  It is nothing but combined GP Experts model. In this model the local experts learn the relationship between the output space and input space.  ... 
doi:10.9790/3021-021021722 fatcat:653cmf3jlnffjjmgr473bjlzwy

Accurate Deep Direct Geo-Localization from Ground Imagery and Phone-Grade GPS

Shaohui Sun, Ramesh Sarukkai, Jack Kwok, Vinay Shet
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
The results show the model can predict quality locations that suffice in real business applications, such as ride-sharing, only using phone-grade GPS.  ...  Unlike classic visual localization or recent PoseNet-like methods that may work well in indoor environments or small-scale outdoor environments, we avoid using a map or an SFM (structure-from-motion) model  ...  Unlike SFM which is often an offline pipeline, Visual SLAM emphasizes real-time capabilities. Mapping and localization are often brought up for discussion at the same time.  ... 
doi:10.1109/cvprw.2018.00148 dblp:conf/cvpr/SunSKS18 fatcat:sanucrz63jhs7c2wdrsbytlowm

Accurate Deep Direct Geo-Localization from Ground Imagery and Phone-Grade GPS [article]

Shaohui Sun, Ramesh Sarukkai, Jack Kwok, Vinay Shet
2018 arXiv   pre-print
The results show the model can predict quality locations that suffice in real business applications, such as ride-sharing, only using phone-grade GPS.  ...  Unlike classic visual localization or recent PoseNet-like methods that may work well in indoor environments or small-scale outdoor environments, we avoid using a map or an SFM (structure-from-motion) model  ...  Unlike SFM which is often an offline pipeline, Visual SLAM emphasizes real-time capabilities. Mapping and localization are often brought up for discussion at the same time.  ... 
arXiv:1804.07470v1 fatcat:rblygithqrh6xbxj3mq76yuwoe

A study on outdoor localization method based on deep learning using model-based received power estimation data of low power wireless tag

Takuto Jikyo, Takahiro Yamanishi, Tomio Kamada, Ryo Nishide, Chikara Ohta, Kenji Oyama, Takenao Ohkawa
2019 IEICE Communications Express  
As existing research, there is a localization method using fingerprint database as learning data in deep learning.  ...  Experimental results showed that an average distance error to GPS data is about 6 m by training DNN using the database and additionally training DNN using actual GPS data.  ...  Localizaiton result in each learning method. localization Average localization Average Standard approach processing time (s) error (m) deviation (m) Matching 7.75×10 2 30.84 16.27 Deep learning  ... 
doi:10.1587/comex.2019gcl0032 fatcat:rtgxd3x25ve6fmfw526hiypuna

Recent Advances in Scaling Up Gaussian Process Predictive Models for Large Spatiotemporal Data [chapter]

Kian Hsiang Low, Jie Chen, Trong Nghia Hoang, Nuo Xu, Patrick Jaillet
2015 Lecture Notes in Computer Science  
The expressive power of Gaussian process (GP) models comes at a cost of poor scalability in the size of the data.  ...  online learning, and nonmyopic active sensing/learning.  ...  For empirical evaluation of GP-Localize with other real-world datasets, refer to [32] .  ... 
doi:10.1007/978-3-319-25138-7_16 fatcat:q2oqq5ijdvdqjd7bkw74q366x4

Sequential Gaussian Processes for Online Learning of Nonstationary Functions [article]

Michael Minyi Zhang, Bianca Dumitrascu, Sinead A. Williamson, Barbara E. Engelhardt
2019 arXiv   pre-print
Many machine learning problems can be framed in the context of estimating functions, and often these are time-dependent functions that are estimated in real-time as observations arrive.  ...  Gaussian processes (GPs) are an attractive choice for modeling real-valued nonlinear functions due to their flexibility and uncertainty quantification.  ...  Typically, SMC methods GP Sparse Local OGP OGP-MOE update a model one sample at a time, but here we update the model using sequential batches.  ... 
arXiv:1905.10003v2 fatcat:lot23jfkzfe5vc7sxrmwv336ni

ShapeMap 3-D: Efficient shape mapping through dense touch and vision [article]

Sudharshan Suresh, Zilin Si, Joshua G. Mangelson, Wenzhen Yuan, Michael Kaess
2022 arXiv   pre-print
Local shape is recovered from tactile images via a learned model trained in simulation.  ...  We demonstrate visuo-tactile mapping in both simulated and real-world experiments, to incrementally build 3-D reconstructions of household objects.  ...  To integrate GelSight tactile images, we recover local shape with a model learned in tactile simulation.  ... 
arXiv:2109.09884v3 fatcat:wnrfc2tutzhitc4gaz4dxb7sla

Learning Non-Stationary Space-Time Models for Environmental Monitoring [article]

Sahil Garg and Amarjeet Singh and Fabio Ramos
2018 arXiv   pre-print
In this paper we propose NOSTILL-GP - NOn-stationary Space TIme variable Latent Length scale GP, a generic non-stationary, spatio-temporal Gaussian Process (GP) model.  ...  We present several strategies, for efficient training of our model, necessary for real-world applicability.  ...  We call this special property of the sparse NOSTILL-GP , Adaptive Local Sparsity. We also use Eq. 5 to model latent GPs for an efficient learning process.  ... 
arXiv:1804.10535v1 fatcat:y43prfbig5fg5kp55caoqfn3aa

Local Gaussian Processes for Efficient Fine-Grained Traffic Speed Prediction

Truc Viet Le, Richard Oentaryo, Siyuan Liu, Hoong Chuin Lau
2017 IEEE Transactions on Big Data  
A local GP corresponding to that cluster can then be trained on the fly to make predictions in real-time. We call this method localization.  ...  In this work, we address their efficiency issues by proposing local GPs to learn from and make predictions for correlated subsets of data.  ...  A local GP corresponding to that cluster can then be trained on the fly to make predictions in real-time. We call this method localization.  ... 
doi:10.1109/tbdata.2016.2620488 fatcat:coqe77syf5fythfka2r7yxwuam

Driven by Vision: Learning Navigation by Visual Localization and Trajectory Prediction

Marius Leordeanu, Iulia Paraicu
2021 Sensors  
Our system learns to predict in real-time vehicle's current location and future trajectory, on a known map, given only the raw video stream and the final destination.  ...  Current published self-driving models improve their performances when using additional GPS information.  ...  In such cases, the GPS could still be working fine. At the same time, the visual-based navigation system, when it works well, it is expected to be real-time and faster than the GPS-based one.  ... 
doi:10.3390/s21030852 pmid:33514019 pmcid:PMC7865778 fatcat:4ezkwpkxfnfaxiijvbsmljlxhq

eFedDNN: Ensemble based Federated Deep Neural Networks for Trajectory Mode Inference [article]

Daniel Opoku Mensah and Godwin Badu-Marfo and Ranwa Al Mallah and Bilal Farooq
2022 arXiv   pre-print
To address this challenge, we use federated learning (FL), a privacy-preserving machine learning technique that aims at collaboratively training a robust global model by accessing users' locally trained  ...  The ensemble method combines the outputs of the different models learned via FL by the users and shows an accuracy that surpasses comparable models reported in the literature.  ...  For the travel mode inference problem, we illustrate that each of the DNN models that serves as the local model in the device of each worker in the FL process needs to learn every sample of the GPS dataset  ... 
arXiv:2205.05756v1 fatcat:okt4v2hkpvenzogyq3jflzcgte
« Previous Showing results 1 — 15 out of 77,958 results