Filters








3,558 Hits in 4.5 sec

Auto-Tuned Sim-to-Real Transfer [article]

Yuqing Du, Olivia Watkins, Trevor Darrell, Pieter Abbeel, Deepak Pathak
2021 arXiv   pre-print
robust to sim-to-real transfer while also not being too conservative.  ...  We evaluate our method on multiple robotic control tasks in both sim-to-sim and sim-to-real transfer, demonstrating significant improvement over naive domain randomization.  ...  in both sim-to-sim and sim-to-real transfer.  ... 
arXiv:2104.07662v2 fatcat:bps2zwfodzcabbvhlf6tp3tlcy

TuneNet: One-Shot Residual Tuning for System Identification and Sim-to-Real Robot Task Transfer [article]

Adam Allevato, Elaine Schaertl Short, Mitch Pryor, Andrea L. Thomaz
2020 arXiv   pre-print
Finally, we show that our method can estimate real-world parameter values, allowing a robot to perform sim-to-real task transfer on a dynamic manipulation task unseen during training.  ...  This paper presents TuneNet, a new machine learning-based method to directly tune the parameters of one model to match another using an *iterative residual tuning* technique.  ...  the reality gap by tuning the physical parameters of a simulator so that it more closely approximates the real world (left), allowing sim-to-real robot skill transfer (right). techniques for predicting  ... 
arXiv:1907.11200v3 fatcat:qecurhtkireqhd76kd3y7hgxs4

A general approach to bridge the reality-gap [article]

Michael Lomnitz, Zigfried Hampel-Arias, Nina Lopatina, Felipe A. Mejia
2020 arXiv   pre-print
However, models trained on these canonical distributions do not readily transfer to real-world ones.  ...  Domain adaptation and transfer learning are often used to breach this "reality gap", though both require a substantial amount of real-world data.  ...  Sim-to-real via sim-to-sim The authors in James et al. [2018] introduced a novel approach to cross the gap between data in real world applications and simulated environments during training of reinforcement  ... 
arXiv:2009.01865v1 fatcat:7fxvqnd77betrbwt4cl2zze3ly

Topic-bridged PLSA for cross-domain text classification

Gui-Rong Xue, Wenyuan Dai, Qiang Yang, Yong Yu
2008 Proceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '08  
A unique advantage of our method is its ability to maximally mine knowledge that can be transferred between domains, resulting in superior performance when compared to other state-of-the-art text classification  ...  By exploiting the common topics between two domains, we transfer knowledge across different domains through a topic-bridge to help the text classification in the target domain.  ...  Two Datasets Generated from SRAA Data Set Traning Data Test Data auto vs aviation sim-auto & sim-aviation real-auto & real-aviation real vs simulated real-aviation & sim-aviation real-auto  ... 
doi:10.1145/1390334.1390441 dblp:conf/sigir/XueDYY08 fatcat:ecidtdmfprhklpnme4jnalredu

On Learning Text Style Transfer with Direct Rewards [article]

Yixin Liu, Graham Neubig, John Wieting
2021 arXiv   pre-print
In most cases, the lack of parallel corpora makes it impossible to directly train supervised models for the text style transfer task.  ...  In particular, we leverage semantic similarity metrics originally used for fine-tuning neural machine translation models to explicitly assess the preservation of content between system outputs and input  ...  , we fine-tuned DIRR-CYCLE on SIM to produce a new model, DIRR w/o FLU.  ... 
arXiv:2010.12771v2 fatcat:fzsazgggjfbylajeuutnjjref4

Offline-Online Learning of Deformation Model for Cable Manipulation with Graph Neural Networks

Changhao Wang, Yuyou Zhang, Xiang Zhang, Zheng Wu, Xinghao Zhu, Shiyu Jin, Te Tang, Masayoshi Tomizuka
2022 IEEE Robotics and Automation Letters  
Then a linear residual model is learned in real-time to bridge the sim-to-real gap.  ...  The online learning and MPC run in a closed-loop manner to robustly accomplish the task.  ...  We compare the performance of the proposed method with three sim-to-real baselines: 1) direct transfer, 2) domain randomization, and 3) fine-tuning the network.  ... 
doi:10.1109/lra.2022.3158376 fatcat:n47afyx3sngvlmywff72sg7fyi

Rare Event Detection Using Disentangled Representation Learning

Ryuhei Hamaguchi, Ken Sakurada, Ryosuke Nakamura
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
A straightforward approach for event detection tasks is to train a detection network from a large-scale dataset in an end-to-end manner.  ...  In order to overcome these difficulties, we propose a novel method to learn disentangled representations from only low-cost negative samples.  ...  From top to bottom, the figures in the columns b and d shows the result of "Under samp.", "Transfer", and "VLAE w/ sim. (ours)", respectively. Results of feature interpolation analysis.  ... 
doi:10.1109/cvpr.2019.00955 dblp:conf/cvpr/HamaguchiSN19 fatcat:5dopbpkugbbw3llx7xfddgg4yq

Adversarial Discriminative Sim-to-real Transfer of Visuo-motor Policies [article]

Fangyi Zhang, Jürgen Leitner, Zongyuan Ge, Michael Milford, Peter Corke
2018 arXiv   pre-print
In this paper, we propose an adversarial discriminative sim-to-real transfer approach to reduce the cost of labelling real data.  ...  The adversarial transfer approach reduced the labelled real data requirement by 50%. Policies can be transferred to real environments with only 93 labelled and 186 unlabelled real images.  ...  for sim-to-real transfer of visuo-motor policies shows the effectiveness of the adversarial discriminative transfer.  ... 
arXiv:1709.05746v2 fatcat:qaauxfe2ozabzl6uimqxf6mlkm

Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors [article]

Artem Molchanov, Tao Chen, Wolfgang Hönig, James A. Preiss, Nora Ayanian, Gaurav S. Sukhatme
2019 arXiv   pre-print
We use reinforcement learning to train policies in simulation that transfer remarkably well to multiple different physical quadrotors.  ...  Quadrotor stabilizing controllers often require careful, model-specific tuning for safe operation.  ...  Sim-to-Sim Verification Before running the policy on real quadrotors, the policy is tested in a different simulator.  ... 
arXiv:1903.04628v2 fatcat:p6hkkdyxdzb5dhq2zxbo7wek5i

Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors

Artem Molchanov, Tao Chen, Wolfgang Honig, James A. Preiss, Nora Ayanian, Gaurav S. Sukhatme
2019 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
The video of our experiments can be found at https:// sites.google.com/view/sim-to-multi-quad.  ...  We use reinforcement learning to train policies in simulation that transfer remarkably well to multiple different physical quadrotors.  ...  Sim-to-Sim Verification Before running the policy on real quadrotors, the policy is tested in a different simulator.  ... 
doi:10.1109/iros40897.2019.8967695 dblp:conf/iros/MolchanovCHPAS19 fatcat:pe5cylsyarca7h3n4say63pzlm

Rare Event Detection using Disentangled Representation Learning [article]

Ryuhei Hamaguchi, Ken Sakurada, Ryosuke Nakamura
2018 arXiv   pre-print
A straightforward approach for event detection tasks is to train a detection network from a large-scale dataset in an end-to-end manner.  ...  In order to overcome these difficulties, we propose a novel method to learn disentangled representations from only low-cost negative samples.  ...  From top to bottom, the figures in the columns b and d shows the result of "Under samp.", "Transfer", and "VLAE w/ sim. (ours)", respectively.  ... 
arXiv:1812.01285v1 fatcat:culwl26pffcvjl4s4c25jk5xdu

CPPF: Towards Robust Category-Level 9D Pose Estimation in the Wild [article]

Yang You, Ruoxi Shi, Weiming Wang, Cewu Lu
2022 arXiv   pre-print
In order to detect objects in the wild, we carefully design our sim-to-real pipeline by training on synthetic point clouds only, unless objects have ambiguous poses in geometry.  ...  Using supervised data of real-world 9D poses is tedious and erroneous, and also fails to generalize to unseen scenarios.  ...  Sim-to-Real Transfer Sim-to-Real is a common strategy in many fields like object reconstruction, pose estimation and reinforcement learning for robots.  ... 
arXiv:2203.03089v2 fatcat:4q6xcvkqfjh27cwun53edcdggq

Successful optimization of reconstruction parameters in structured illumination microscopy - a practical guide [article]

Christian Karras, Maria Smedh, Ronny Foerster, Hendrik Deschout, Julia Fernandez-Rodriguez, Rainer Heintzmann
2018 bioRxiv   pre-print
The impact of the different reconstruction parameters in super-resolution structured illumination microscopy (SIM) onto artifacts is carefully analyzed.  ...  They comprise the Wiener filter parameter, an apodization function, zero-frequency suppression and modifications of the optical transfer function.  ...  To this aim, the simulated (real-valued) transfer function is for instance simply raised to an exponent slightly different than one (c.f. supplementary material, sect. 4).  ... 
doi:10.1101/402115 fatcat:ojrmncfguzgijoxbhyhcx2h3gm

Civil Rephrases Of Toxic Texts With Self-Supervised Transformers [article]

Leo Laugier, John Pavlopoulos, Jeffrey Sorensen, Lucas Dixon
2021 arXiv   pre-print
CAE-T5 employs a pre-trained text-to-text transformer, which is fine tuned with a denoising and cyclic auto-encoder loss.  ...  style transfer systems which we compare with using several scoring systems and human evaluation.  ...  Second, their approach involves collaborative classifiers to penalize generation when the attribute is not transferred, while we train end-to-end with a denoising auto-encoder.  ... 
arXiv:2102.05456v2 fatcat:6ky4ma4d6zcqrc6ryueqbkesfq

Lane Detection in Low-light Conditions Using an Efficient Data Enhancement : Light Conditions Style Transfer [article]

Tong Liu, Zhaowei Chen, Yi Yang, Zehao Wu, Haowei Li
2020 arXiv   pre-print
Our solution consists of three parts: the proposed SIM-CycleGAN, light conditions style transfer and lane detection network.  ...  In this paper, we propose a style-transfer-based data enhancement method, which uses Generative Adversarial Networks (GANs) to generate images in low-light conditions, that increases the environmental  ...  Image Amounts to Generate We fine tune the number of generated images for our lane detection model.  ... 
arXiv:2002.01177v2 fatcat:vc2p7fdsf5dk7ozr5w2jnjzqnq
« Previous Showing results 1 — 15 out of 3,558 results