4,943 Hits in 4.5 sec

Gated recurrent units viewed through the lens of continuous time dynamical systems [article]

Ian D. Jordan, Piotr Aleksander Sokol, Il Memming Park
2019 arXiv   pre-print
Gated recurrent units (GRUs) are specialized memory elements for building recurrent neural networks.  ...  Using a continuous time analysis, we gain intuition on the inner workings of GRU networks. We restrict our presentation to low dimensions to allow for a comprehensive visualization.  ...  Underlying Continuous Time System of Gated Recurrent Units The GRU uses two internal gating variables: the update gate z t which protects the d-dimensional hidden state h t ∈ R d and the reset gate r t  ... 
arXiv:1906.01005v1 fatcat:gnuomdtltvhu3iqzzzzot6jo6y

API Call-Based Malware Classification Using Recurrent Neural Networks

Chen Li, Junjun Zheng
2021 Journal of Cyber Security and Mobility  
In this paper, we propose a malware classification model with RNN, especially the long short-term memory (LSTM) and the gated recurrent unit (GRU), to classify variants of malware by using long-sequences  ...  Recurrent neural networks (RNNs) is one of the most versatile approaches to process time series data, which can be used to API call-based Malware calssification.  ...  Gated Recurrent Unit Different from the LSTM cell, the structure of GRU is relatively simple. Figure 3 shows the basic structure of a GRU model.  ... 
doi:10.13052/jcsm2245-1439.1036 fatcat:fl4z4thenvcphfeejp6b32zqfe

Birhythmic Analog Circuit Maze: A Nonlinear Neurostimulation Testbed

Ian D. Jordan, Il Memming Park
2020 Entropy  
The circuit spontaneously oscillates stably on two periods as an instantiation of a 3-dimensional continuous-time gated recurrent neural network.  ...  In order to develop next-generation neural stimulators that can intelligently learn to induce attractor transitions, we require a platform to test the efficacy of such systems.  ...  In the following section, we derive the system from the general continuous-time dynamical system underlying the gated recurrent unit (GRU), a commonly used recurrent neural network architecture [33, 34  ... 
doi:10.3390/e22050537 pmid:33286310 pmcid:PMC7517031 fatcat:ghud3nhlevfitklsv2d6r5ve2m

Drawing and Recognizing Chinese Characters with Recurrent Neural Network [article]

Xu-Yao Zhang and Fei Yin and Yan-Ming Zhang and Cheng-Lin Liu and Yoshua Bengio
2016 arXiv   pre-print
With the RNN system (combining an LSTM and GRU), state-of-the-art performance can be achieved on the ICDAR-2013 competition database.  ...  Experimental results verify the effectiveness of using RNNs as both generative and discriminative models for the tasks of drawing and recognizing Chinese characters.  ...  ACKNOWLEDGMENTS The authors thank the developers of Theano [44] , [45] for providing such a good deep learning platform.  ... 
arXiv:1606.06539v1 fatcat:ny7tle5szrfq3pacz5ejwe3dza

Prefrontal cortex as a meta-reinforcement learning system

Jane X. Wang, Zeb Kurth-Nelson, Dharshan Kumaran, Dhruva Tirumala, Hubert Soyer, Joel Z. Leibo, Demis Hassabis, Matthew Botvinick
2018 Nature Neuroscience  
A prevalent view 24 is that the role of the striatum within this cortico-basal ganglia-thalamocortical loop is to regulate the dynamics of PFC by gating the flow of information into PFC circuits.  ...  Instead, learning reflects the activation dynamics of the recurrent network.  ...  The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the funding agencies.  ... 
doi:10.1038/s41593-018-0147-8 pmid:29760527 fatcat:exhfpleg4zgkrjzw7uy5wegwqa

Prefrontal cortex as a meta-reinforcement learning system [article]

Jane X Wang, Zeb Kurth-Nelson, Dharshan Kumaran, Dhruva Tirumala, Hubert Soyer, Joel Z Leibo, Demis Hassabis, Matthew Botvinick
2018 bioRxiv   pre-print
Here, the dopamine system trains another part of the brain, the prefrontal cortex, to operate as its own free-standing learning system.  ...  and rewards by modulating the strength of synaptic connections between neurons.  ...  The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the funding agencies.  ... 
doi:10.1101/295964 fatcat:rfugghouzrdnro2hagmnvvt7ni

Tricks from Deep Learning [article]

Atılım Güneş Baydin and Barak A. Pearlmutter and Jeffrey Mark Siskind
2016 arXiv   pre-print
Many of these methods, viewed through the lens of algorithmic differentiation (AD), can be seen as either addressing issues with the gradient itself, or finding ways of achieving increased efficiency using  ...  Two of these are discussed in further mathematical detail: a way to dramatically reduce the size of the tape when performing reverse-mode AD on a (theoretically) time-reversible process like an ODE integrator  ...  Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsors.  ... 
arXiv:1611.03777v1 fatcat:ryzokp4b7jcz7onhqiruofwhfe

State estimation with limited sensors – A deep learning based approach [article]

Yash Kumar, Pranav Bahl, Souvik Chakraborty
2021 arXiv   pre-print
The proposed model structure consists of the recurrent cell to pass information from different time steps enabling utilization of this information to recover the full state.  ...  Such approaches, in general, use measurement data of one-time instance. However, oftentimes data available from sensors is sequential and ignoring it results in information loss.  ...  SC acknowledges the financial support of the I-Hub Foundation for Cobotics (IHFC) and seed grant provided through Faculty initiation grant, IIT Delhi.  ... 
arXiv:2101.11513v2 fatcat:cn76zb6ya5dxjatpule2w4ed2i

A Novel pseudo viewpoint based Holoscopic 3D Micro-gesture Recognition

Yi Liu, Shuang Yang, Hongying Meng, Mohammad Rafiq Swash, Shiguang Shan
2020 Companion Publication of the 2020 International Conference on Multimodal Interaction  
Then, a new recognition framework based on 3D DenseNet and Bi-GRU networks is proposed to learn the dynamic patterns of different micro-gestures based on the representation of the pseudo view points.  ...  Finally, we perform a thorough comparison on the related benchmark, which demonstrates the effectiveness of our method and also reports a new state of the art performance.  ...  They introduced the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) based networks and achieved 65.4% and 69.2% respectively.  ... 
doi:10.1145/3395035.3425249 fatcat:xgdikbzcjjevhkqhrbxcitsxxy

A Multimode Optical Imaging System for Preclinical Applications In Vivo: Technology Development, Multiscale Imaging, and Chemotherapy Assessment

Jae Youn Hwang, Sebastian Wachsmann-Hogiu, V. Krishnan Ramanujan, Julia Ljubimova, Zeev Gross, Harry B. Gray, Lali K. Medina-Kauwe, Daniel L. Farkas
2011 Molecular Imaging and Biology  
modalities to examine dynamic characteristics of cells and tissues.  ...  Results-The approach offers a comprehensive imaging platform for kinetic, quantitative, and environmental analysis of highly relevant information, with micro-to-macroscopic resolution.  ...  Mark Gaon for help developing and testing the gated anesthesia instrument. Some of this work was done in partial fulfillment of Ph.D  ... 
doi:10.1007/s11307-011-0517-z pmid:21874388 pmcid:PMC3487699 fatcat:k4br3dpmcjbt5e7vpqqi5smrby

Freudian and Newtonian Recurrent Cell for Sequential Recommendation [article]

Hoyeop Lee, Jinbae Im, Chang Ouk Kim, Sehee Chung
2021 arXiv   pre-print
The predominant sequential recommendation models are based on natural language processing models, such as the gated recurrent unit, that embed items in some defined space and grasp the user's long-term  ...  FaNC initiates a new direction of sequential recommendations at the convergence of psychoanalysis and recommender systems.  ...  Specifically, the recurrent layer has the same architecture as the gated recurrent unit (GRU) [4] because the GRU outperforms other recurrent cells for sequential recommendation [8] .  ... 
arXiv:2102.07645v1 fatcat:paac3mp65jd2jbt6r77kqczj4q

A low-resolution vision sensor

D G Whitehead, I Mitchell, P V Mellor
1984 Journal of Physics E Scientific Instruments  
A low-resolution camera system based on a 64k dynamic RAM chip is described. The camera is capable of providing a picture resolution of up to 256 x 128 pixels.  ...  A fast microprocessor is used to collect the picture data and to provide a preprocessing facility and a flexible means of access to the data.  ...  Acknowledgments This work has been funded by the Science and Engineering Research Council under contract number GRlBl4586.3 and GRlBl5940.2.  ... 
doi:10.1088/0022-3735/17/8/008 fatcat:ztqbhtrjk5cbzh5fzjxx6hqkra

DANA: Distributed numerical and adaptive modelling framework

Nicolas P. Rougier, Jérémy Fix
2012 Network  
DANA is a python framework ( whose computational paradigm is grounded on the notion of a unit that is essentially a set of time dependent values varying under the influence of other  ...  , recurrent neural networks, kernel-based image processing, etc.). similar to a logical unit able to carry out quite complex computations while in the case of the blue brain project, the concept of a neuron  ...  Acknowledgment Declaration of interest The authors report no conflicts of interest.  ... 
doi:10.3109/0954898x.2012.721573 pmid:22994650 fatcat:gk354f4ddnd2xixymtp4q2ebye

Latent Factors and Dynamics in Motor Cortex and Their Application to Brain–Machine Interfaces

Chethan Pandarinath, K. Cora Ames, Abigail A. Russo, Ali Farshchian, Lee E. Miller, Eva L. Dyer, Jonathan C. Kao
2018 Journal of Neuroscience  
Here we discuss this dynamical systems perspective and how it is reshaping our understanding of the motor cortices.  ...  Recent experimental results suggest that the dynamical properties of these networks are critical to movement planning and execution.  ...  Acknowledgements The authors would like to thank Steven Chase, Chandramouli Chandrasekaran, Juan Gallego  ... 
doi:10.1523/jneurosci.1669-18.2018 pmid:30381431 pmcid:PMC6209846 fatcat:etwyvltypncp7miookcmy5rqfy

Active gaze stabilization

Martin Lesmana, Axel Landgren, Per-Erik Forssén, Dinesh K. Pai
2014 Proceedings of the 2014 Indian Conference on Computer Vision Graphics and Image Processing - ICVGIP '14  
We describe a system for active stabilization of cameras mounted on highly dynamic robots.  ...  We evaluated the performance quantitatively using another high speed robot to generate repeatable sequences of large and fast movements that a gaze stabilization system can attempt to counteract.  ...  The lens used provides a 74 • horizontal viewing angle.  ... 
doi:10.1145/2683483.2683565 dblp:conf/icvgip/LesmanaLFP14 fatcat:sx5yhvyxj5duhbrb7gha4dgs3u
« Previous Showing results 1 — 15 out of 4,943 results