257 Hits in 2.9 sec

Muesli: Combining Improvements in Policy Optimization [article]

Matteo Hessel, Ivo Danihelka, Fabio Viola, Arthur Guez, Simon Schmitt, Laurent Sifre, Theophane Weber, David Silver, Hado van Hasselt
2022 arXiv   pre-print
We propose a novel policy update that combines regularized policy optimization with model learning as an auxiliary loss.  ...  Notably, Muesli does so without using deep search: it acts directly with a policy network and has computation speed comparable to model-free baselines.  ...  Joseph Modayil improved the paper by wise comments and advice.  ... 
arXiv:2104.06159v2 fatcat:4jafvxdd55f4tdj2vgt647gsxe

Model-Value Inconsistency as a Signal for Epistemic Uncertainty [article]

Angelos Filos, Eszter Vértes, Zita Marinho, Gregory Farquhar, Diana Borsa, Abram Friesen, Feryal Behbahani, Tom Schaul, André Barreto, Simon Osindero
2022 arXiv   pre-print
We provide empirical evidence in both tabular and function approximation settings from pixels that self-inconsistency is useful (i) as a signal for exploration, (ii) for acting safely under distribution  ...  Unlike prior work which estimates uncertainty by training an ensemble of many models and/or value functions, this approach requires only the single model and value function which are already being learned in  ...  We used the mean of IVE components in place of the learned policy for acting (µ-IVE( 5 )), and combined the mean with the self-inconsistency signal for acting optimistically in the face of uncertainty  ... 
arXiv:2112.04153v3 fatcat:rt5kzvijxbfxzldv376ednxex4

Self-Consistent Models and Values [article]

Gregory Farquhar, Kate Baumli, Zita Marinho, Angelos Filos, Matteo Hessel, Hado van Hasselt, David Silver
2021 arXiv   pre-print
In particular, models enable planning, i.e. using more computation to improve value functions or policies, without requiring additional environment interactions.  ...  We propose multiple self-consistency updates, evaluate these in both tabular and function approximation settings, and find that, with appropriate choices, self-consistency helps both policy evaluation  ...  We also evaluated the same self-consistency update in a small experiment on self-play 9x9 Go in combination with a Muesli-MPO baseline, with results shown in Figure 3c .  ... 
arXiv:2110.12840v1 fatcat:5ott7uqvavhodldt6nimv2ussu

fAn exploratory study using graphic design to communicate consumer benefits on food packaging

Hendrik N.J. Schifferstein, Mailin Lemke, Alie de Boer
2021 Food Quality and Preference  
Our findings suggest that consumers can handle multiple packaging messages, but finding an optimal configuration remains a design challenge.  ...  For three products (orange juice, muesli bar, plain yogurt) we created three consistent packaging designs communicating a single benefit through all three mediums, which was either a [1] health, [2] environmental  ...  Acknowledgements The authors are indebted to the editors and anonymous reviewers who helped to improve the quality of this paper.  ... 
doi:10.1016/j.foodqual.2021.104458 fatcat:bb6feiteijbvvpth7xjxi63vwi

Generalized Data Distribution Iteration [article]

Jiajun Fan, Changnan Xiao
2022 arXiv   pre-print
optimization.  ...  In this paper, we try to tackle these two challenges simultaneously.  ...  MUESLI Muesli (Hessel et al., 2021) proposed a novel policy update that combines regularized policy optimization with model learning as an auxiliary loss.  ... 
arXiv:2206.03192v4 fatcat:7qtr7ms2hzgodhtzj6oneof2ba

Growing the Business of Whole Grain in the Australian Market: A 6-Year Impact Assessment

Curtain, Locke, Grafenauer
2020 Nutrients  
Three-quarters (78% and 74%) of the eligible breakfast cereals and bread products were registered with the Code in 2019, followed by 62% of grain-based muesli bars.  ...  Reporting included breakfast cereals, bread products, crispbreads, crackers, rice/corn cakes, rice, pasta, noodles, couscous, other grains (e.g., quinoa, buckwheat, freekeh), and grain-based muesli bars  ...  Acknowledgments: Thanks to Nikki Lancaster, Student Dietitian from the University of Wollongong, NSW, who was involved in data collection and analysis for price comparisons.  ... 
doi:10.3390/nu12020313 pmid:31991603 pmcid:PMC7071175 fatcat:grpa4xdbrrcajkwjyu4et7dpsi

GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning [article]

Jiajun Fan, Changnan Xiao, Yue Huang
2022 arXiv   pre-print
Deep Q Network (DQN) firstly kicked the door of deep reinforcement learning (DRL) via combining deep learning (DL) with reinforcement learning (RL), which has noticed that the distribution of the acquired  ...  From this new perspective, we extend the basic paradigm of RL called the Generalized Policy Iteration (GPI) into a more generalized version, which is called the Generalized Data Distribution Iteration  ...  Theorem 2 (Second Order Optimization with Superior Improvement).  ... 
arXiv:2106.06232v6 fatcat:hsxgrwj2kzcytnisynzlcjltni

Hybrid CPU–GPU execution support in the skeleton programming framework SkePU

Tomas Öhberg, August Ernstsson, Christoph Kessler
2019 Journal of Supercomputing  
In this paper, we present a hybrid execution backend for the skeleton programming framework SkePU.  ...  Acknowledgements This work has been partly funded by EU H2020 project EXA2PRO (801015), by SeRC (http://www.e-scien, and by the Swedish National Graduate School in Computer Science (CUGS).  ...  distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in  ... 
doi:10.1007/s11227-019-02824-7 fatcat:oiifchkxdzdqrmwpxz5m6adroa

Structured Data Access Annotations for Massively Parallel Computations [chapter]

Marco Aldinucci, Sonia Campa, Peter Kilpatrick, Massimo Torquati
2013 Lecture Notes in Computer Science  
In most cases, cost models are used to drive the refactoring process.  ...  We show how sample use case applications/kernels may be optimized and discuss preliminary experiments with FastFlow assessing the theoretical results.  ...  transpose of P 1 and sp 1 , cm 2 define generic splitting and combining policies not influencing P 1 .  ... 
doi:10.1007/978-3-642-36949-0_42 fatcat:fwrd3j45vffirfzyfsudknhpjq

Algorithmic skeletons meeting grids

Marco Danelutto, Marco Aldinucci
2006 Parallel Computing  
In this work, we discuss an extension of the set of principles that should guide the future design and development of skeletal programming systems, as defined by Cole in his "pragmatic manifesto" paper  ...  Dynamicity handling (7) is not supported at all in Muesli. Gorlatch's group work was more related on data parallel skeleton optimizations, actually [39, 40] .  ...  No support for dynamicity handling (7) is provided in eSkel, however. Muesli is basically a C++ library built on top of MPI.  ... 
doi:10.1016/j.parco.2006.04.001 fatcat:v5fbdwdk6zcf3m2s3fm3ysfdwa

The effectiveness of healthy meals at work on reaction time, mood and dietary intake: a randomised cross-over study in daytime and shift workers at an university hospital

Eva Leedo, Anne Marie Beck, Arne Astrup, Anne D. Lassen
2017 British Journal of Nutrition  
Providing healthy meals, snacks and water during working hours seems to be an effective way of improving employees' dietary intake.  ...  Score (P=0·017), whereas the only dietary component that significantly improved was water intake (P=0·034), when compared with the control period.  ...  authors thank Katrine Rask and Annette Vedelspang for the practical help during the study period, Heidi Anker for processing the dietary data, Michael Allerup Nielsen for securing funding and support in  ... 
doi:10.1017/s000711451700191x pmid:28820084 fatcat:hhhmvpb5ofh5xkc2xywuarhoqu

Co-design of Distributed Systems Using Skeleton and Autonomic Management Abstractions [chapter]

M. Aldinucci, M. Danelutto, P. Kilpatrick
2009 Lecture Notes in Computer Science  
In particular, we demonstrate how restricted parallel/distributed patterns (or skeletons) may be efficiently managed by rule-based autonomic managers.  ...  The following core (abstract) skeleton set has been defined (in slightly different forms) in a large number of skeleton frameworks, including P3L [4] , Muesli [14] , Lithium/muskel [3, 8] , SkeTo [  ...  The sustained evolution in parallel and distributed architectures makes application development even harder, as technological improvements and architectural model changes must be catered for.  ... 
doi:10.1007/978-3-642-00955-6_46 fatcat:ik77qzcqrzfqtn5t4lyiptdg5i

Food-Based Dietary Guidelines in Austria

I. Elmadfa, H. Freisling
2007 Annals of Nutrition and Metabolism  
Minor differences arise in the translation of this NBDG into food servings.  ...  In Austria, first steps have already been taken and several promising FBDG have been introduced to the consumer.  ...  FBDG have proved to be an appropriate health policy measure for the improvement of nutrition.  ... 
doi:10.1159/000103561 fatcat:nqltgi54jbeyfaqvfz2ssp2i74

Healthiness of Food and Beverages for Sale at Two Public Hospitals in New South Wales, Australia

Carrie Tsai, Erika Svensen, Victoria Flood, Yasmine Probst, Kathryn Reilly, Stephen Corbett, Jason Wu
2018 Nutrients  
These findings highlight the need for ongoing tracking to inform whether the revised guidelines are leading to improved food environments in health facilities.  ...  These findings highlight the need for ongoing tracking to inform whether the revised guidelines are leading to improved food environments in health facilities.  ...  Acknowledgments: The University of Sydney, Australia (Healthy Sydney University Iniative) provided funding support for this study including the costs to publish in open access.  ... 
doi:10.3390/nu10020216 pmid:29462881 pmcid:PMC5852792 fatcat:rumpagvjyvh2fccbgqriihdjgi

Fastflow: High-Level and Efficient Streaming on Multicore [chapter]

Marco Aldinucci, Marco Danelutto, Peter Kilpatrick, Massimo Torquati
2017 Programming multi-core and many-core computing systems  
support has been-in some cases, e.g. in Muesli and Muskel-only a later addition.  ...  typically used in combination with a farm body to model Divide&Conquer computations).  ... 
doi:10.1002/9781119332015.ch13 fatcat:x4fvha2dofdxpoq5xjow2eaa4q
« Previous Showing results 1 — 15 out of 257 results