Filters








304 Hits in 8.3 sec

Weight-Sharing Neural Architecture Search: A Battle to Shrink the Optimization Gap [article]

Lingxi Xie, Xin Chen, Kaifeng Bi, Longhui Wei, Yuhui Xu, Zhengsu Chen, Lanfei Wang, An Xiao, Jianlong Chang, Xiaopeng Zhang, Qi Tian
2020 arXiv   pre-print
This paper provides a literature review on NAS, in particular the weight-sharing methods, and points out that the major challenge comes from the optimization gap between the super-network and the sub-architectures  ...  To alleviate the burden, weight-sharing methods were proposed in which exponentially many architectures share weights in the same super-network, and the costly training procedure is performed only once  ...  ACKNOWLEDGMENTS The authors would like to thank the colleagues in Huawei Noah's Ark Lab, in particular, Prof. Tong Zhang, Dr. Wei Zhang, Dr.  ... 
arXiv:2008.01475v2 fatcat:bsmbepvg3bbezeim6izpi5a2ey

A Survey of Monte Carlo Tree Search Methods

Cameron B. Browne, Edward Powley, Daniel Whitehouse, Simon M. Lucas, Peter I. Cowling, Philipp Rohlfshagen, Stephen Tavener, Diego Perez, Spyridon Samothrakis, Simon Colton
2012 IEEE Transactions on Computational Intelligence and AI in Games  
Monte Carlo Tree Search (MCTS) is a recently proposed search method that combines the precision of tree search with the generality of random sampling.  ...  This paper is a survey of the literature to date, intended to provide a snapshot of the state of the art after the first five years of MCTS research.  ...  The value of a chance node is the sum of its children weighted by their probabilities, otherwise the search is identical to max n .  ... 
doi:10.1109/tciaig.2012.2186810 fatcat:z2o6xdkkjvhf3k6hybw65o5r44

Deep Learning in Mobile and Wireless Networking: A Survey [article]

Chaoyun Zhang, Paul Patras, Hamed Haddadi
2019 arXiv   pre-print
In this paper we bridge the gap between deep learning and mobile and wireless networking research, by presenting a comprehensive survey of the crossovers between the two areas.  ...  One potential solution is to resort to advanced machine learning techniques to help managing the rise in data volumes and algorithm-driven applications.  ...  The AutoML platform 2 provides a first solution to this problem, by employing progressive neural architecture search [110] . This task, however, remains costly.  ... 
arXiv:1803.04311v3 fatcat:awuvyviarvbr5kd5ilqndpfsde

Deep Neural Networks for Optimal Team Composition

Anna Sapienza, Palash Goyal, Emilio Ferrara
2019 Frontiers in Big Data  
We then use such framing to design a recommendation system to suggest new teammates based on a modified deep neural autoencoder and we demonstrate its state-of-the-art recommendation performance.  ...  We generate a directed co-play network, whose links' weights depict the effect of teammates on players' performance.  ...  For deep neural network based models, we use ReLU as the activation function and choose the neural network structure by an informal search over a set of architectures.  ... 
doi:10.3389/fdata.2019.00014 pmid:33693337 pmcid:PMC7931874 fatcat:ibhamjrpefbvbkecwa5jeaa3ie

A perspective on the future of massively parallel computing

Predrag T. Tosic
2004 Proceedings of the first conference on computing frontiers on Computing frontiers - CF'04  
A great variety of parallel computation models has been proposed and studied, and different parallel and distributed architectures designed as some possible ways of harnessing parallelism and improving  ...  Massively parallel connectionist models such as artificial neural networks (ANNs) and cellular automata (CA) have been primarily studied in domain-specific contexts, namely, learning and complex dynamics  ...  Acknowledgments The author is greatly indebted to Gul Agha, Tom Anastasio, Alfred Hubler and Sylvian Ray, all at University of Illinois.  ... 
doi:10.1145/977091.977160 dblp:conf/cf/Tosic04 fatcat:5jhbmelpzndxno4447xbzs6sk4

InterpoNet, a Brain Inspired Neural Network for Optical Flow Dense Interpolation

Shay Zweig, Lior Wolf
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Sparse-to-dense interpolation for optical flow is a fundamental phase in the pipeline of most of the leading optical flow estimation algorithms.  ...  We propose a new data-driven sparse-to-dense interpolation algorithm based on a fully convolutional network.  ...  Acknowledgments This research is supported by the Intel Collaborative Research Institute for Computational Intelligence (ICRI-CI) and by the Israeli Ministry of Science, Technology, and Space.  ... 
doi:10.1109/cvpr.2017.674 dblp:conf/cvpr/ZweigW17 fatcat:pup4gsguqvb2nagebnzing7xua

Backward Chaining [chapter]

2013 Encyclopedia of Operations Research and Management Science  
An approach to reasoning in which an inference engine endeavors to find a value for an overall goal by recursively finding values for subgoals.  ...  At any point in the recursion, the effort of finding a value for the immediate goal involves examining rule conclusions to identify those rules that could possibly establish a value for that goal.  ...  The polynomial complexity of the path-following method for LP follows immediately from the fact that each Newton step shrinks the primal-dual gap by ð1 À a= ffiffiffi ffi m p Þ, where a > 0 is independent  ... 
doi:10.1007/978-1-4419-1153-7_200968 fatcat:wsdsqf2wxrckvpoq4yh5f7rsqm

Deep learning for small and big data in psychiatry

Georgia Koppe, Andreas Meyer-Lindenberg, Daniel Durstewitz
2020 Neuropsychopharmacology  
This power comes at a cost, the need for large training (and test) samples to infer the (sometimes over millions of) model parameters.  ...  We review how machine learning approaches compare to more traditional statistical hypothesis-driven approaches, how their complexity relates to the need of large sample sizes, and what we can do to optimally  ...  From the discussion in the previous section one may deduce that for such models a huge amount of data is needed to battle the bias-variance trade-off.  ... 
doi:10.1038/s41386-020-0767-z pmid:32668442 fatcat:c4aon35pf5c4ro3qikvw7qbgsi

Can We `Feel' the Temperature of Knowledge? Modelling Scientific Popularity Dynamics via Thermodynamics [article]

Luoyi Fu, Dongrui Lu, Qi Li, Xinbing Wang, Chenghu Zhou
2020 arXiv   pre-print
The spike is especially obvious when there appears a single non-trivial novel research focus or merging in topic structure.  ...  Our analysis of representative topics with size ranging from 1000 to over 30000 articles reveals that the key to flourishing is topics' ability in accumulating useful information for future knowledge generation  ...  -If y i > y j , add a directed weighted edge from a j to a i of weight 2·∑ px A pxx m(m−1) . The new edge means "a j virtually cites a i ".  ... 
arXiv:2007.13270v1 fatcat:qybvondcsfg2fh7gb2goxfwvnm

InterpoNet, A brain inspired neural network for optical flow dense interpolation [article]

Shay Zweig, Lior Wolf
2017 arXiv   pre-print
Sparse-to-dense interpolation for optical flow is a fundamental phase in the pipeline of most of the leading optical flow estimation algorithms.  ...  We propose a new data-driven sparse-to-dense interpolation algorithm based on a fully convolutional network.  ...  While the anatomic resemblance of such models to the cortex is appealing, in reality, they are unfolded to a feedforward network with shared weights.  ... 
arXiv:1611.09803v3 fatcat:fkxyy76efjdbhignucz4wmmrrq

The rise of machine learning for detection and classification of malware: Research developments, trends and challenges

Daniel Gibert, Carles Mateu, Jordi Planes
2020 Journal of Network and Computer Applications  
A B S T R A C T The struggle between security analysts and malware developers is a never-ending battle with the complexity of malware changing as quickly as innovation grows.  ...  The survey helps researchers to have an understanding of the malware detection field and of the new developments and directions of research explored by the scientific community to tackle the problem.  ...  This research article has received a grant (2019 call) from the University of Lleida Language Institute to review the English.  ... 
doi:10.1016/j.jnca.2019.102526 fatcat:3bf6afjqpnb53eoeghfxjeaus4

The neural basis of cognitive development: a constructivist manifesto

S R Quartz, T J Sejnowski
1997 Behavioral and Brain Sciences  
architecture.  ...  According to "neural constructivism," the representational features of cortex are built from the dynamic interaction between neural growth mechanisms and environmentally derived neural activity.  ...  the space in which the nondynamical learning algorithm searches for the optimal or suboptimal parameters.  ... 
pmid:10097006 fatcat:ammov422u5dy7aouvbm5ezhg2e

The neural basis of cognitive development: A constructivist manifesto

Steven R. Quartz, Terrence J. Sejnowski
1997 Behavioral and Brain Sciences  
architecture.  ...  According to "neural constructivism," the representational features of cortex are built from the dynamic interaction between neural growth mechanisms and environmentally derived neural activity.  ...  the space in which the nondynamical learning algorithm searches for the optimal or suboptimal parameters.  ... 
doi:10.1017/s0140525x97001581 fatcat:2x57otwpxrcwpklsixe3sukshu

Indexical Cities: Articulating Personal Models of Urban Preference with Geotagged Data [article]

Diana Alvarez-Marin, Karla Saldana Ochoa
2020 arXiv   pre-print
How to assess the potential of liking a city or a neighborhood before ever having been there.  ...  The concept of urban quality has until now pertained to global city ranking, where cities are evaluated under a grid of given parameters, or either to empirical and sociological approaches, often constrained  ...  The stochastic gradient descent is used to optimize the network loss function, starting with a learning rate of 0.1, and halving the rate every ten epochs.  ... 
arXiv:2001.10615v1 fatcat:th2h5uzptrfengwh3sjob75jeu

Text Modular Networks: Learning to Decompose Tasks in the Language of Existing Models [article]

Tushar Khot and Daniel Khashabi and Kyle Richardson and Peter Clark and Ashish Sabharwal
2021 arXiv   pre-print
We use this framework to build ModularQA, a system that can answer multi-hop reasoning questions by decomposing them into sub-questions answerable by a neural factoid single-span QA model and a symbolic  ...  These sub-questions and answers provide a faithful natural language explanation of the model's reasoning.  ...  Acknowledgements We thank the Aristo team at AI2 for helpful input, Beaker team for their support with experiments, Dirk Groeneveld for providing the output of the Quark system for evaluation, and Jonathan  ... 
arXiv:2009.00751v2 fatcat:gd6534kconcrlo2rko2vpmg2em
« Previous Showing results 1 — 15 out of 304 results