Filters








2,839 Hits in 7.3 sec

The Microsoft 2016 Conversational Speech Recognition System

W. Xiong, J. Droppo, X. Huang, F. Seide, M. Seltzer, A. Stolcke, D. Yu, G. Zweig
2017 arXiv   pre-print
We describe Microsoft's conversational speech recognition system, in which we combine recent developments in neural-network-based acoustic and language modeling to advance the state of the art on the Switchboard  ...  Inspired by machine learning ensemble techniques, the system uses a range of convolutional and recurrent neural networks.  ...  Chen from CUED for valuable assistance with the CUED-RNNLM toolkit, and ICSI for compute and data resources.  ... 
arXiv:1609.03528v2 fatcat:n7oti5vqwndfpdircvuxaytnm4

The microsoft 2016 conversational speech recognition system

W. Xiong, J. Droppo, X. Huang, F. Seide, M. Seltzer, A. Stolcke, D. Yu, G. Zweig
2017 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
We describe Microsoft's conversational speech recognition system, in which we combine recent developments in neural-network-based acoustic and language modeling to advance the state of the art on the Switchboard  ...  Inspired by machine learning ensemble techniques, the system uses a range of convolutional and recurrent neural networks.  ...  Chen from CUED for valuable assistance with the CUED-RNNLM toolkit, and ICSI for compute and data resources.  ... 
doi:10.1109/icassp.2017.7953159 dblp:conf/icassp/XiongDHSSSYZ17 fatcat:7nqwwcpaprcppdgtgqgvjugfle

Component reuse methodology for multi-clock Data-Flow parallel embedded Systems

Anne Marie Chana, Patrice Quinton, Steven Derrien
2014 ARIMA  
The work presented here is an extension of previous work. We illustrate our method on a simplified WCDMA system.  ...  Systems on Chip (SoC) based on reused components have become an absolute necessity to embedded systems companies that want to remain competitive.  ...  The calculation of the periods of our components appears as a transposition, in the domain of recurrence equations, of the balance equations of SDF.  ... 
doi:10.46298/arima.1979 fatcat:bi6gyfa27jd5vatza752ubttae

Leveraging machine translation for cross-lingual fine-grained cyberbullying classification amongst pre-adolescents

Kanishk Verma, Maja Popović, Alexandros Poulis, Yelena Cherkasova, Cathal Ó hÓbáin, Angela Mazzone, Tijana Milosevic, Brian Davis
2022 Natural Language Engineering  
Similar to face-to-face bullying, cyberbullying can be captured formally using the Routine Activities Model (RAM) whereby the potential victim and bully are brought into proximity of one another via the  ...  Although the impact of the COVID-19 (SARS-CoV-2) restrictions on the online presence of minors has yet to be fully grasped, studies have reported that 44% of pre-adolescents have encountered more cyberbullying  ...  children: A rights-based approach to fighting bullying.  ... 
doi:10.1017/s1351324922000341 fatcat:jisirukk5nfhxjb4fouz56grra

A brain basis of dynamical intelligence for AI and computational neuroscience [article]

Joseph D. Monaco, Kanaka Rajan, Grace M. Hwang
2021 arXiv   pre-print
This article was inspired by our symposium on dynamical neuroscience and machine learning at the 6th Annual US/NIH BRAIN Initiative Investigators Meeting.  ...  To motivate a brain basis of neural computation, we present a dynamical view of intelligence from which we elaborate concepts of sparsity in network structure, temporal dynamics, and interactive learning  ...  The analysis of mechanistic couplings over time is the domain of dynamical systems theory.  ... 
arXiv:2105.07284v2 fatcat:ble5h45pk5fczn72dwco2m3rkm

CoBERL: Contrastive BERT for Reinforcement Learning [article]

Andrea Banino, Adrià Puidomenech Badia, Jacob Walker, Tim Scholtes, Jovana Mitrovic, Charles Blundell
2022 arXiv   pre-print
CoBERL enables efficient, robust learning from pixels across a wide range of domains.  ...  We use bidirectional masked prediction in combination with a generalization of recent contrastive methods to learn better representations for transformers in RL, without the need of hand engineered data  ...  Third, while COBERL uses an extension of RELIC (Mitrovic et al., 2021) to the time domain and operates on the inputs and outputs of the transformer, M-CURL uses CPC (Oord et al., 2018) with a momentum  ... 
arXiv:2107.05431v2 fatcat:dn3kj5bzybbnbhemxxwr4hrt3a

Oscillations and Filtering Networks Support Flexible Routing of Information

Thomas Akam, Dimitri M. Kullmann
2010 Neuron  
We show that switching one of several convergent pathways from an asynchronous to an oscillatory state allows accurate selective transmission of population-coded information, which can be extracted even  ...  Task-dependent changes in the power and interregion coherence of network oscillations suggest that such oscillations play a role in signal routing.  ...  ACKNOWLEDGMENTS This work was supported by the Wellcome Trust and the European Research Council. We are grateful to P. Latham for helpful discussion and comments on the manuscript.  ... 
doi:10.1016/j.neuron.2010.06.019 pmid:20670837 pmcid:PMC3125699 fatcat:u6lfp4pmebbxtkh2vkram4dafm

Tracking recurrence of correlation structure in neuronal recordings

Samuel A. Neymotin, Zoe N. Talbot, Jeeyune Q. Jung, André A. Fenton, William W. Lytton
2017 Journal of Neuroscience Methods  
PCo allows intuitive, visual assessment of temporal recurrence in correlation structure directly in the high dimensionality dataset, allowing for immediate assessment of relevant dynamics at a single site  ...  h i g h l i g h t s • PCo, a multiscale method, determines the recurrence of neural correlation structure. • PCo operates at multiple temporal and spatial scales without dimensional reduction. • PCo detects  ...  Acknowledgements Supported by grants from the Simons Foundation (294388), and National Institutes of Health: R01EB022903; R01MH084038; R01MH099128; R01MH086638; R42NS064474; U01EB017695.  ... 
doi:10.1016/j.jneumeth.2016.10.009 pmid:27746231 pmcid:PMC5266613 fatcat:loylojhjrvfsvbaoqlkohqkjpi

Correcting Diacritics and Typos with a ByT5 Transformer Model

Lukas Stankevičius, Mantas Lukoševičius, Jurgita Kapočiūtė-Dzikienė, Monika Briedienė, Tomas Krilavičius
2022 Applied Sciences  
Our simultaneous diacritics restoration and typos correction approach reaches >94% alpha-word accuracy on the 13 languages.  ...  For a comparison, we perform diacritics restoration on benchmark datasets of 12 languages, with the addition of Lithuanian.  ...  There is, typically, the so-called warm-up period in the beginning to level discrepancies between previous parameters and new domain updates.  ... 
doi:10.3390/app12052636 fatcat:relakw7ovff63oxsudtj2oyyk4

Achieving Human Parity in Conversational Speech Recognition [article]

W. Xiong, J. Droppo, X. Huang, F. Seide, M. Seltzer, A. Stolcke, D. Yu, G. Zweig
2017 arXiv   pre-print
recurrent neural network language modeling approaches, and a systematic use of system combination.  ...  The key to our system's performance is the use of various convolutional and LSTM acoustic model architectures, combined with a novel spatial smoothing method and lattice-free MMI acoustic training, multiple  ...  Acknowledgments We thank Arul Menezes for access to the Microsoft transcription pipeline; Chris Basoglu, Amit Agarwal and Marko Radmilac for their invaluable assistance with CNTK; Jinyu Li and Partha Parthasarathy  ... 
arXiv:1610.05256v2 fatcat:rkx4tebqena23okl6mi6bnibu4

PsychRNN: An Accessible and Flexible Python Package for Training Recurrent Neural Network Models on Cognitive Tasks

Daniel B. Ehrlich, Jasmine T. Stone, David Brandfonbrener, Alexander Atanasov, John D. Murray
2020 eNeuro  
First, the task of interest is defined, and a recurrent neural network (RNN) model is trained to perform the task, optionally with neurobiologically informed constraints on the network.  ...  The training backend is based on TensorFlow and is readily extensible for researchers with TensorFlow knowledge to develop projects with additional customization.  ...  Simulator One limitation of specifying RNN networks in the TensorFlow language is that to run a network, the inputs, outputs, and computation need to take place within the TensorFlow framework, which can  ... 
doi:10.1523/eneuro.0427-20.2020 fatcat:3wm3rfwhcrfhxkck2at5e7cety

Stock2Vec: A Hybrid Deep Learning Framework for Stock Market Prediction with Representation Learning and Temporal Convolutional Network [article]

Xing Wang, Yijun Wang, Bin Weng, Aleksandr Vinel
2020 arXiv   pre-print
We have proposed to develop a global hybrid deep learning framework to predict the daily prices in the stock market.  ...  Evaluated on S&P 500, our hybrid framework integrates both advantages and achieves better performance on the stock price prediction task than several popular benchmarked models.  ...  [47] , speech recognition [48] , natural language processing [49] , and extensions to autoregressive time series forecasting [50, 51] in recent years.  ... 
arXiv:2010.01197v1 fatcat:qub27gaeabez3bzooraoddzji4

Correcting diacritics and typos with a ByT5 transformer model [article]

Lukas Stankevičius, Mantas Lukoševičius, Jurgita Kapočiūtė-Dzikienė, Monika Briedienė, Tomas Krilavičius
2022 arXiv   pre-print
Our simultaneous diacritics restoration and typos correction approach reaches > 94% alpha-word accuracy on the 13 languages.  ...  For a comparison, we perform diacritics restoration on benchmark datasets of 12 languages, with the addition of Lithuanian.  ...  Conflicts of Interest: The authors declare no conflict of interest.  ... 
arXiv:2201.13242v2 fatcat:o5rn53acofbg3p52mfeyucgngu

High-performance parallel implicit CFD

William D Gropp, Dinesh K Kaushik, David E Keyes, Barry F Smith
2001 Parallel Computing  
We document both claims from our experience with an unstructured grid CFD code that is typical of the state of the practice at NASA.  ...  special attention to layout and access ordering of data.  ...  Kyle Anderson, formerly of the NASA Langley Research Center, and to Dimitri Mavriplis of ICASE for collaborations leading up to the work presented herein.  ... 
doi:10.1016/s0167-8191(00)00075-2 fatcat:boiejxon5vhe5mfh2bumonwhti

Deep Predictive Learning in Neocortex and Pulvinar

Randall C. O'Reilly, Jacob L. Russin, Maryam Zolfaghar, John Rohrlich
2021 Journal of Cognitive Neuroscience  
We implemented these mechanisms in a large-scale model of the visual system and found that the simulated inferotemporal pathway learns to systematically categorize 3-D objects according to invariant shape  ...  Specifically, numerous weak projections into the pulvinar nucleus of the thalamus generate top–down predictions, and sparse driver inputs from lower areas supply the actual outcome, originating in Layer  ...  Acknowledgments We thank Dean Wyatte, Tom Hazy, Seth Herd, Kai Krueger, Tim Curran, David Sheinberg, Lew Harvey, Jessica Mollick, Will Chapman, Helene Devillez, and the rest of the CCN Lab for many helpful  ... 
doi:10.1162/jocn_a_01708 pmid:34428793 fatcat:tbyyilt5b5axtorr3p4wmuesoe
« Previous Showing results 1 — 15 out of 2,839 results