Filters








157 Hits in 6.1 sec

DARTS-: Robustly Stepping out of Performance Collapse Without Indicators [article]

Xiangxiang Chu, Xiaoxing Wang, Bo Zhang, Shun Lu, Xiaolin Wei, Junchi Yan
2021 arXiv   pre-print
Various indicators such as Hessian eigenvalues are proposed as a signal to stop searching before the performance collapses.  ...  Despite the fast development of differentiable architecture search (DARTS), it suffers from long-standing performance instability, which extremely limits its application.  ...  STEPPING OUT OF THE PERFORMANCE COLLAPSE To distinguish the two roles, we introduce an auxiliary skip connection between every two nodes in a cell, see Fig. 1 (b).  ... 
arXiv:2009.01027v2 fatcat:fvskev6pybbnjg6qllmdjtp7fi

β-DARTS: Beta-Decay Regularization for Differentiable Architecture Search [article]

Peng Ye, Baopu Li, Yikang Li, Tao Chen, Jiayuan Fan, Wanli Ouyang
2022 arXiv   pre-print
However, they suffer from two main issues, the weak robustness to the performance collapse and the poor generalization ability of the searched architectures.  ...  Among them, differential NAS approaches such as DARTS, have gained popularity for the search efficiency.  ...  Related Works Robustness of DARTS As DARTS is known to chronically suffer from the performance collapse issue caused by the domination of parameter-free operators, lots of works have dedicated to resolving  ... 
arXiv:2203.01665v2 fatcat:3q3daee5lvhs5elcvu6vujzb3e

Noisy Differentiable Architecture Search [article]

Xiangxiang Chu, Bo Zhang
2021 arXiv   pre-print
However, it largely suffers from the well-known performance collapse issue due to the aggregation of skip connections.  ...  We perform extensive experiments across various search spaces, datasets, and tasks, where we robustly achieve state-of-the-art results.  ...  Stepping out of the performance collapse by noise Based on the above analysis, we propose NoisyDARTS to step out of the performance collapse.  ... 
arXiv:2005.03566v3 fatcat:soqlordgljbpxgjqmh423qhkii

HexEx

Max Lyon, David Bommes, Leif Kobbelt
2016 ACM Transactions on Graphics  
In a postprocessing step, we let certain pairs of darts cancel each other, counteracting the effect of flipped regions of the parametrization.  ...  Abstract State-of-the-art hex meshing algorithms consist of three steps: Frame-field design, parametrization generation, and mesh extraction.  ...  Some of the models are courtesy of the AIM@SHAPE Repository.  ... 
doi:10.1145/2897824.2925976 fatcat:zyfs2rsklrd6lhmbfbof6fnjh4

eIF2B activator prevents neurological defects caused by a chronic Integrated Stress Response [article]

Yao Liang Wong, Lauren LeBon, Ana M. Basso, Kathy L. Kohlhaas, Arthur L. Nikkel, Holly M. Robb, Diana L. Donnelly-Roberts, Janani Prakash, Andrew M. Swensen, Nimrod D. Rubinstein, Swathi Krishnan, Fiona E. McAllister (+8 others)
2018 bioRxiv   pre-print
AbstractThe Integrated Stress Response (ISR) attenuates the rate of protein synthesis while inducing expression of stress proteins in cells.  ...  of mutant eIF2B complex in vivo, abrogating the maladaptive stress response.  ...  All steps were performed on dry ice prior to the addition of 350 µL of RTL Buffer + 40 µM DTT.  ... 
doi:10.1101/462820 fatcat:s2hndao7f5bkdplbqixvlkdovi

eIF2B activator prevents neurological defects caused by a chronic integrated stress response

Yao Liang Wong, Lauren LeBon, Ana M Basso, Kathy L Kohlhaas, Arthur L Nikkel, Holly M Robb, Diana L Donnelly-Roberts, Janani Prakash, Andrew M Swensen, Nimrod D Rubinstein, Swathi Krishnan, Fiona E McAllister (+10 others)
2019 eLife  
The integrated stress response (ISR) attenuates the rate of protein synthesis while inducing expression of stress proteins in cells.  ...  We show that introduction of a human VWM mutation into mice leads to persistent ISR induction in the central nervous system. ISR activation precedes myelin loss and development of motor deficits.  ...  All steps were performed on dry ice prior to the addition of 350 mL of RTL Buffer + 40 mM DTT.  ... 
doi:10.7554/elife.42940 pmid:30624206 pmcid:PMC6326728 fatcat:lsomc64f4bhq5cmygfhjpqejsi

Joint Inference of Kinematic and Force Trajectories with Visuo-Tactile Sensing [article]

Alexander Lambert, Mustafa Mukadam, Balakumar Sundaralingam, Nathan Ratliff, Byron Boots, Dieter Fox
2019 arXiv   pre-print
of the contact itself.  ...  To perform complex tasks, robots must be able to interact with and manipulate their surroundings.  ...  Right: visualization of the tracked system in DART [21] , with the observed pointcloud marked in dark grey. (after the last iSAM2 step has been performed).  ... 
arXiv:1903.03699v1 fatcat:qu6ufpnvu5bzlfi3l7buii65ba

Simultaneous Constraints on the Star Formation History and Nucleosynthesis of Sculptor dSph [article]

Mithi A. C. de los Reyes, Evan N. Kirby, Alexander P. Ji, Evan H. Nuñez
2021 arXiv   pre-print
Our best-fit model indicates that Sculptor dSph has an ancient SFH, consistent with previous estimates from deep photometry.  ...  We use a one-zone chemical evolution model to fit the overall abundance patterns of α elements (which probe the core-collapse supernovae that occur shortly after star formation), s-process elements (which  ...  The final time step, t final is not a free parameter. Rather, it is the last time step before the galaxy runs out of gas.  ... 
arXiv:2110.01690v2 fatcat:g6oeavgxc5bhdilpulyvcd24di

The mitochondrial genome of Paraspadella gotoi is highly reduced and reveals that chaetognaths are a sister group to protostomes

K. G. Helfenbein, H. M. Fourcade, R. G. Vanjani, J. L. Boore
2004 Proceedings of the National Academy of Sciences of the United States of America  
A dart (>) marks the last nucleotide for each gene and indicates the direction of transcription. 1,802 positions are parsimony informative.  ...  Our analysis recovered a single most parsimonious tree of 16,267 steps (Fig. 4) that robustly supports the Chaetognatha-represented here by P. gotoi-as the sister group to the protostomes included in  ... 
doi:10.1073/pnas.0400941101 pmid:15249679 pmcid:PMC489987 fatcat:wz4txix3jfgl7jhyhzl5s6qtvq

Error-Bounded and Feature Preserving Surface Remeshing with Minimal Angle Improvement

Kaimo Hu, Dong-Ming Yan, David Bommes, Pierre Alliez, Bedrich Benes
2017 IEEE Transactions on Visualization and Computer Graphics  
free of bad elements that would hamper the desired application.  ...  More specifically, we iteratively perform carefully prioritized local operators, whenever they do not violate the approximation error bound and improve the mesh otherwise.  ...  ACKNOWLEDGMENTS We wish to thank Mario Botsch for providing the source code and results of their method, and Pascal Frey for providing the MMGS software. This work was partially supported  ... 
doi:10.1109/tvcg.2016.2632720 pmid:28114021 fatcat:wh5conpzzbeptk7bujymmwpge4

Error-Bounded and Feature Preserving Surface Remeshing with Minimal Angle Improvement [article]

Kaimo Hu, Dong-Ming Yan, David Bommes, Pierre Alliez, Bedrich Benes
2016 arXiv   pre-print
More specifically, we iteratively perform carefully prioritized local operators, whenever they do not violate the approximation error bound and improve the mesh otherwise.  ...  free of bad elements that would hamper the desired application.  ...  ACKNOWLEDGMENTS We wish to thank Mario Botsch for providing the source code and results of their method, and Pascal Frey for providing the MMGS software. This work was partially supported  ... 
arXiv:1611.02147v1 fatcat:mex5vsvagfbl5oivygm2bdgcni

Guided Feature Transformation (GFT): A Neural Language Grounding Module for Embodied Agents [article]

Haonan Yu, Xiaochen Lian, Haichao Zhang, Wei Xu
2018 arXiv   pre-print
Recently there has been a rising interest in training agents, embodied in virtual environments, to perform language-directed tasks by deep reinforcement learning.  ...  We also release XWorld3D, an easy-to-customize 3D environment that can potentially be modified to evaluate a variety of embodied agents.  ...  Overall, GFT performs multi-step concept detection with language-dependent filters at each step.  ... 
arXiv:1805.08329v2 fatcat:gyhr2b3wlbfo5fyv7727erghhu

Electroencephalographic connectivity measures predict learning of a motor sequencing task

Jennifer Wu, Franziska Knapp, Steven C. Cramer, Ramesh Srinivasan
2018 Journal of Neurophysiology  
These results indicate that connectivity in the resting state is a better predictor of consolidated learning of motor skills.  ...  A resting-state measure of whole brain coherence with primary motor cortex (M1) at baseline robustly predicted subsequent motor improvement (validated R 2 ϭ 0.55) and motor learning (validated R 2 ϭ 0.68  ...  Cramer has served as a consultant for Dart Neuroscience, Micro-Transponder, and Toyama. No conflicts of interest, financial or otherwise, are declared by the authors.  ... 
doi:10.1152/jn.00580.2017 pmid:29093171 fatcat:qiucnejmsjchfdwilprty4jur4

Science, society and biosafety of a field trial with transgenic biofuel poplars

Rebecca Van Acker, Véronique Storme, Geert Goeminne, Bart Ivens, René Custers, Dirk Aerts, Wim Soetaert, John Ralph, Nicholas Santoro, Jean-Charles Leplé, Gilles Pilate, Wout Boerjan
2011 BMC Proceedings  
Recent applications of high-throughput sequencing and genotyping technologies to natural populations of forest Figure 1 (abstract I2) The distribution of Douglas-fir and loblolly pine in North America.  ...  A long history of quantitative genetic experimentation has established the genetic basis for many traits, which are likely adaptive since many of them are also correlated with environmental heterogeneity  ...  The availability of the QMI to rapidly assess the potential risk generated by transgenic plants in characterized ecosystems, is a useful tool to assess the releasing or not of the permission of cultivation  ... 
doi:10.1186/1753-6561-5-s7-i23 fatcat:ulkxqgdupfdktkhsx2kba7pcn4

A Survey on Policy Search for Robotics

Marc Peter Deisenroth
2011 Foundations and Trends in Robotics  
It is well suited for robotics as it can cope with high-dimensional state and action spaces, one of the main challenges in robot learning.  ...  We review recent successes of both model-free and model-based policy search in robot learning. Model-free policy search is a general approach to learn policies based on sampled trajectories.  ...  (c) Movement learned with the eNAC algorithm after 300 roll-outs. The robot learned to hit the ball robustly.  ... 
doi:10.1561/2300000021 fatcat:bfka3vw4njgrhjiz5kbpey42nq
« Previous Showing results 1 — 15 out of 157 results