Filters








241,717 Hits in 5.2 sec

Adaptive Procedural Task Generation for Hard-Exploration Problems [article]

Kuan Fang, Yuke Zhu, Silvio Savarese, Li Fei-Fei
2021 arXiv   pre-print
We introduce Adaptive Procedural Task Generation (APT-Gen), an approach to progressively generate a sequence of tasks as curricula to facilitate reinforcement learning in hard-exploration problems.  ...  At the heart of our approach, a task generator learns to create tasks from a parameterized task space via a black-box procedural generation module.  ...  We would like to thank Roberto Martín-Martín, Austin Narcomey, Sriram Somasundaram, Fei Xia, and Danfei Xu for feedback on an early draft of the paper.  ... 
arXiv:2007.00350v3 fatcat:325qus6ovjhonf2fheip6sahei

Automatic Curriculum Learning For Deep RL: A Short Survey [article]

Rémy Portelas, Cédric Colas, Lilian Weng, Katja Hofmann, Pierre-Yves Oudeyer
2020 arXiv   pre-print
In recent years, they have been used to improve sample efficiency and asymptotic performance, to organize exploration, to encourage generalization or to solve sparse reward problems, among others.  ...  adapted to their capacities.  ...  A more general -and arguably more powerful-approach is to leverage parametric Procedural Content Genera-tion (PCG) techniques [Risi and Togelius, 2019] to generate rich task spaces.  ... 
arXiv:2003.04664v2 fatcat:lhire3htmnenfetx2ry4furgyy

Learning Not to Learn: Nature versus Nurture in Silico [article]

Robert Tjarko Lange, Henning Sprekeler
2022 arXiv   pre-print
Further analysis reveals that non-adaptive behaviors are not only optimal for aspects of the environment that are stable across individuals, but also in situations where an adaptation to the environment  ...  'hard-coded' behavior.  ...  Acknowledgments We thank Nir Moneta for many initial discussions.  ... 
arXiv:2010.04466v3 fatcat:eo5gmen7zfbxhaao6gz563rbhi

Towards Effective Context for Meta-Reinforcement Learning: an Approach based on Contrastive Learning [article]

Haotian Fu, Hongyao Tang, Jianye Hao, Chen Chen, Xidong Feng, Dong Li, Wulong Liu
2020 arXiv   pre-print
By conditioning on an effective context, Meta-RL policies can easily generalize to new tasks within a few adaptation steps.  ...  Context, the embedding of previous collected trajectories, is a powerful construct for Meta-Reinforcement Learning (Meta-RL) algorithms.  ...  Acknowledgements We thank Chenjia Bai, Qianli Shen, and Weixun Wang for useful discussions and suggestions.  ... 
arXiv:2009.13891v3 fatcat:6mpsw4qkuzgcpfv2hgtcskyze4

Self-Guided Adaptation: Progressive Representation Alignment for Domain Adaptive Object Detection [article]

Zongxian Li, Qixiang Ye, Chong Zhang, Jingjing Liu, Shijian Lu and Yonghong Tian
2020 arXiv   pre-print
The core of SGA is to calculate "hardness" factors for sample pairs indicating domain distance in a kernel space.  ...  Indicated by hardness factors, Self-Guided Progressive Sampling (SPS) is implemented in an "easy-to-hard" way during model adaptation.  ...  Model Analysis For further exploring the effectiveness of proposed approaches, we analyze both training procedures and detection results from following aspects, which are performed over tasks: Pascal VOC  ... 
arXiv:2003.08777v2 fatcat:6xj52frxqrerrnx2lmtzh7cm2i

Hybridized TABU-BFO Algorithmin Grid Scheduling

Devi Priya, Joshua Samuel Raj, V. Vasudeven
2013 International Journal of Computer Applications  
Tabu search is a heuristic procedure which uses its adaptive memory structures in order to find optimal solution in grid scheduling.  ...  Task Scheduling is an important issue in Grid Environment of multiprocessors system. The problem of scheduling a set of dependent and independent Task in a distributed system is considered.  ...  One of the main issue in task Scheduling is the NP-Hard problem. Thus to avoid such a problem in the grid environment Tabu search concept is introduced.  ... 
doi:10.5120/10514-5484 fatcat:4ike5w67m5b6pdrn6k47ak66yy

PARSSSE: AN ADAPTIVE PARALLEL STATE SPACE SEARCH ENGINE

YANHUA SUN, GENGBIN ZHENG, PRITISH JETLEY, LAXMIKANT V. KALE
2011 Parallel Processing Letters  
Solving such problems is generally NP-hard, so that a brute-force approach to state space search must be employed.  ...  Moreover, we tackle the problem of programmer productivity by incorporating these techniques into a general search engine framework designed to solve a broad class of state space search problems.  ...  Acknowledgments This work was supported in part by NSF grant OCI-0725070 for Blue Waters deployment and NSF grant ITR-HECURA-0833188, by the Institute for Advanced Computing Applications and Technologies  ... 
doi:10.1142/s0129626411000242 fatcat:nqdtfz46qzal5mv3kxpzypirui

A Review of Methodologies for Natural-Language-Facilitated Human-Robot Cooperation [article]

Rui Liu, Xiaoli Zhang
2017 arXiv   pre-print
In this review, a comprehensive summary about methodologies for NLC is presented.  ...  NLC research includes three main research focuses: NL instruction understanding, NL-based execution plan generation, and knowledge-world mapping.  ...  Typical tackled problems include using MLN to generate a flexible machine-executable plan from human NL instructions for autonomous industrial task execution [28] , NL-based cooperation in uncertain environments  ... 
arXiv:1701.08756v3 fatcat:gquzgq3w6bcf5ih4aubchyyulm

Automated Curriculum Learning for Turn-level Spoken Language Understanding with Weak Supervision [article]

Hao Lang, Wen Wang
2019 arXiv   pre-print
Furthermore, considering the diversity of problem complexity, we explore automated curriculum learning (CL) for weak supervision to accelerate exploration and learning.  ...  We employ randomized beam search with memory augmentation (RBSMA) to solve complicated problems for which long promising trajectories are usually difficult to explore.  ...  We propose two techniques for better exploration and generalization: (1) RBSMA for complicated problems with long programs, (2) automated CL for weakly supervised learning for dealing with the diversity  ... 
arXiv:1906.04291v1 fatcat:jro2hpn2gjatrlimyqcwv7xlne

Self-explanatory user interfaces by model-driven engineering

Alfonso García Frey
2010 Proceedings of the 2nd ACM SIGCHI symposium on Engineering interactive computing systems - EICS '10  
We explore modeldriven engineering to understand why and how this approach can lead us to overcome shortcomings of UI quality successfully.  ...  Modern User Interfaces (UI) must deal with the increasing complexity of applications as well as new features such as the capacity of UIs to be dynamically adapted to the context of use.  ...  Even many long-time users never master common procedures [6] and in other cases, users must work hard to figure out each feature or screen [6] . The problem is greater for Plastic UIs [5, 19] .  ... 
doi:10.1145/1822018.1822076 dblp:conf/eics/Frey10 fatcat:olfwqledm5arfewwchqov2pjsm

DAWSON: A Domain Adaptive Few Shot Generation Framework [article]

Weixin Liang, Zixuan Liu, Can Liu
2020 arXiv   pre-print
To this end, we propose DAWSON, a Domain Adaptive FewShot Generation FrameworkFor GANs based on meta-learning.  ...  Training a Generative Adversarial Networks (GAN) for a new domain from scratch requires an enormous amount of training data and days of training time.  ...  Acknowledgement This work was done as a course project for CS 236 Deep Generative Models at Stanford University.  ... 
arXiv:2001.00576v1 fatcat:wjud4apyubdlzohgmiaw5qb4mm

Collective Decision Making under Incomplete Knowledge: Possible and Necessary Solutions

Jérôme Lang
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
Most solution concepts in collective decision making are defined assuming complete knowledge of individuals' preferences and of the mechanism used for aggregating them.  ...  A more general -and arguably more powerful-approach is to leverage parametric Procedural Content Generation (PCG) techniques [Risi and Togelius, 2019] to generate rich task spaces.  ...  ACL mechanisms are used with a particular goal in mind (e.g. organizing exploration, solving hard tasks, etc. § 3). It controls a particular element of task MDPs (e.g.  ... 
doi:10.24963/ijcai.2020/671 dblp:conf/ijcai/PortelasCWHO20 fatcat:nqkxitg6yzbzvor2c3bi6rib5q

A Survey on Transfer Learning for Multiagent Reinforcement Learning Systems

Felipe Leno Da Silva, Anna Helena Reali Costa
2019 The Journal of Artificial Intelligence Research  
We define a taxonomy of solutions for the general knowledge reuse problem, providing a comprehensive discussion of recent progress on knowledge reuse in Multiagent Systems (MAS) and of techniques for knowledge  ...  Multiagent Reinforcement Learning (RL) solves complex tasks that require coordination with other agents through autonomous exploration of the environment.  ...  Taylor for the collaboration in previous versions of this survey.  ... 
doi:10.1613/jair.1.11396 fatcat:mn4gw6oh5zgszl6l53fgesei5i

A Brief Look at Generalization in Visual Meta-Reinforcement Learning [article]

Safa Alver, Doina Precup
2020 arXiv   pre-print
In this paper, we assess the generalization performance of these algorithms by leveraging high-dimensional, procedurally generated environments.  ...  We also observe that scalability to high-dimensional tasks with sparse rewards remains a significant problem among many of the current meta-reinforcement learning algorithms.  ...  These works use procedural content generation to evaluate generalization in regular RL algorithms.  ... 
arXiv:2006.07262v3 fatcat:7zkmbze76vb7rjl4vd4judrqyu

A new approach to comparing problem solving, flexibility and innovation

Alice M.I. Auersperg, Gyula K. Gajdon, Auguste M.P. von Bayern
2012 Communicative & Integrative Biology  
One major reason for this is that it is very hard to find universally applicable paradigms that can be used to investigate the same cognitive ability or 'general intelligence' in several species.  ...  Potentially revealing profiles could be obtained from examining species differences in how novel experimental (extractive foraging) tasks are explored and approached, how solutions are discovered and which  ...  It highlights how different performance in problem solving task are not always symptomatic of species differences in cognitive ability or general intelligence.  ... 
doi:10.4161/cib.18787 pmid:22808317 pmcid:PMC3376048 fatcat:2ffmgrevfjd37akaxdqy5hwpmu
« Previous Showing results 1 — 15 out of 241,717 results