Filters








66,528 Hits in 4.6 sec

Bearing fault diagnosis based on Natural Adaptive Moment Estimation algorithm and improved octave convolution

Mei Ying Qiao, Xia Xia Tang, Jian Ke Shi, Jian Yi Lan
2020 IEEE Access  
The low-frequency component needs to be up-sampled during information updating. → is the weight parameter for converting highfrequency data to low-frequency data, and for down-sampling high-frequency components  ...  In Fig 2, the green arrows indicate information updates, and the red arrows indicate information exchange between the two frequencies.  ... 
doi:10.1109/access.2020.3034281 fatcat:br5u26uv3fcrdopncoc22r3e54

Feature-Frequency–Adaptive On-line Training for Fast and Accurate Natural Language Processing

Xu Sun, Wenjie Li, Houfeng Wang, Qin Lu
2014 Computational Linguistics  
It is based on the core idea that higher frequency features should have a learning rate that decays faster.  ...  To reach this target, we present a new training method, featurefrequency-adaptive on-line training, for fast and accurate training of natural language processing systems.  ...  vector in the ADF learning algorithm, which is decaying over the time t and with different decaying rates based on feature frequency information.  ... 
doi:10.1162/coli_a_00193 fatcat:xnweb3j4ebfz5aibvlm5bo2nwu

Block-Online Guided Source Separation [article]

Shota Horiguchi, Yusuke Fujita, Kenji Nagamatsu
2020 arXiv   pre-print
Using the context enables the algorithm to estimate time-frequency masks accurately only from one iteration of optimization for each block, and its latency does not depend on the utterance length but predetermined  ...  GSS is a speech separation method that uses diarization information to update parameters of the generative model of observation signals.  ...  The diarization information helps to make the model free from the frequency permutation problem because it is frequency-independent.  ... 
arXiv:2011.07791v1 fatcat:cdr3tm6udjeqdkvouhdqnjvgd4

Neural ODEs for Informative Missingness in Multivariate Time Series [article]

Mansura Habiba, Barak A. Pearlmutter
2020 arXiv   pre-print
Informative missingness is unavoidable in the digital processing of continuous time series, where the value for one or more observations at different time points are missing.  ...  These datasets often consist of multiple variables, and often there are missing values for one or many of these variables.  ...  Here, (f) is the update function that constructs the dynamics of hidden as well as derivative of the cell state and call the ODE solver to update all the parameters for optimizer at once.  ... 
arXiv:2005.10693v1 fatcat:7yuqk23syjgplgp4txceufv7xm

Trip-Oriented Model Predictive Energy Management Strategy for Plug-in Hybrid Electric Vehicles

Zhenzhen Lei, Dongye Sun, Junjun Liu, Daqi Chen, Yonggang Liu, Zheng Chen
2019 IEEE Access  
First, an exponentially varied model for the velocity prediction is established, of which the tunable decay coefficient is regulated by the supported vector machine (SVM).  ...  Moreover, the typical DP algorithm is leveraged to solve the nonlinear rolling optimization problem for minimizing the operating cost in a receding horizon.  ...  Second, the MPC algorithm should consider the updating frequency of the traffic information, hence, simulations with different updating frequencies are conducted to find the optimal frequency.  ... 
doi:10.1109/access.2019.2933015 fatcat:iksuubbi4zbirbwkrmohyejeua

Optimization-based design and implementation of multidimensional zero-phase IIR filters

D. Gorinevsky, S. Boyd
2006 IEEE Transactions on Circuits and Systems I Fundamental Theory and Applications  
Transfer function magnitude specifications, convergence speed requirements for the iterative implementation, and spatial decay of the filter impulse response (which defines the boundary condition influence  ...  A rigorous optimization-based design of the filter is considered.  ...  The problem statement included the iteration convergence/spatial decay condition (60) and the optimality criterion (69). In the design, a 32 32 grid was used for the frequency-dependent functions.  ... 
doi:10.1109/tcsi.2005.856048 fatcat:nck7iledhzbrxf7dgleeqj4tgm

Importance Sampling based Exploration in Q Learning [article]

Vijay Kumar, Mort Webster
2021 arXiv   pre-print
ADP approximates the optimal value function by adaptively sampling both action and state space.  ...  We propose a novel approach for selecting actions using importance sampling weighted by the value function approximation in continuous decision spaces to address this dilemma.  ...  Epsilon Greedy and Epsilon Decay require similar computation time; Epsilon Decay requires slightly more time because the increasing frequency of exploit samples requires more optimizations of the Q-function  ... 
arXiv:2107.00602v1 fatcat:4oghpyy4nvd6xgzllge2mcni6m

addHessian: Combining quasi-Newton method with first-order method for neural network training

Sota Yasuda, S. Indrapriyadarsini, Hiroshi Ninomiya, Takeshi Kamio, Hideki Asai
2022 Nonlinear Theory and Its Applications IEICE  
In this paper, we propose a new learning algorithm for training neural networks by combining first-order and second-order methods.  ...  On the other hand, second-order methods have shown to have better performance and faster convergence despite their high computational cost by incorporating the curvature information.  ...  Several studies have been focusing on effective stochastic optimization methods. Choice of the optimization algorithm used for training the NN plays an important role.  ... 
doi:10.1587/nolta.13.361 fatcat:wzdhkiepz5adjmuwxexdqpo4oy

Dynamic Advisor-Based Ensemble (dynABE): Case study in stock trend prediction of critical metal companies

Zhengyang Dong, Stefan Cristian Gherghina
2019 PLoS ONE  
an online update strategy we developed. dynABE is able to adapt to price pattern changes of the market during the active trading period robustly, without needing to retrain the entire model.  ...  In this research, we create a novel framework for stock prediction, Dynamic Advisor-Based Ensemble (dynABE). dynABE explores domain-specific areas based on the companies of interest, diversifies the feature  ...  The author would also like to thank Asuka Saito from Nogizaka46 for her motivations when bottlenecks were met during the research process.  ... 
doi:10.1371/journal.pone.0212487 pmid:30794608 pmcid:PMC6386270 fatcat:gjmnz2j42nfixhebng74lghvrq

Attentional and Contextual Priors in Sound Perception

Michael Wolmetz, Mounya Elhilali, Joel Snyder
2016 PLoS ONE  
Further analyses showed that listeners updated their contextual priors rapidly and optimally, given the changing acoustic frequency statistics inherent in the paradigm.  ...  (c) Prior probability alters sensitivity similarly across stimulus classes and for informatively and uninformatively cued tones.  ...  The authors thank research assistants Katherine Simeon and Nicha Apichitsopa for their valuable suggestions and contributions to data collection.  ... 
doi:10.1371/journal.pone.0149635 pmid:26882228 pmcid:PMC4755547 fatcat:yrmg2liflvcsjfyocy4hhn3hcq

Laser Spot Detection Using Robust Dictionary Construction and Update

Zhihua Wang, Yongri Piao, Minglu Jin
2015 Journal of information and communication convergence engineering  
The decay rate  is used against noises; a small  value is sensitive to noises, and a large one cannot adapt to a low frequency of background changes.  ...  Since the use of video sequences as a dictionary is sensitive to noise, we use information from multiple frames for ensuring robustness.  ... 
doi:10.6109/jicce.2015.13.1.042 fatcat:2um5z3r4n5blzdvdo7iwrhmbue

PACL: Piecewise Arc Cotangent Decay Learning Rate For Deep Neural Network Training

Haixu Yang, Jihong Liu, Hongwei Sun, Henggui Zhang
2020 IEEE Access  
INDEX TERMS Deep neural networks, learning rate schedulers, arc cotangent, optimization.  ...  Deep neural networks (DNNs) are currently the best-performing method for many classification problems.  ...  ESTIMATE FREQUENCY OF LEARNING RATE TRANSFORMATION In the practical application of PACL, div is introduced to represent the update frequency of the learning rate in each epoch.  ... 
doi:10.1109/access.2020.3002884 fatcat:ylf2yyp6cnfhbiuqyqstz7gxxq

Predicting the Daily Covariance Matrix for S&P 100 Stocks using Intraday Data - But Which Frequency to Use?

Michiel De Pooter, Martin P.E. Martens, Dick J. C. van Dijk
2005 Social Science Research Network  
We focus on the issue of determining the optimal sampling frequency as judged by the performance of these portfolios.  ...  Bias-correction procedures, based on combining low-frequency and high-frequency covariance matrix estimates and on the addition of leads and lags do not substantially affect the optimal sampling frequency  ...  Results Optimal decay rates This pattern implies that the update V t−1,h in (14) is given more weight when it is measured, presumably more accurately, at higher sampling frequencies.  ... 
doi:10.2139/ssrn.829545 fatcat:obzguf4x3fcwhlqm4m7wzq3ejy

New Measurement of the Bs Mixing Phase at CDF [article]

Gavril Giurgiu for the CDF Collaboration
2010 arXiv   pre-print
The CDF collaboration presents an updated measurement of the CP-violating parameter beta_s and of the decay width difference Delta Gamma_s using approximately 6500 Bs to J/Psi Phi decays collected by the  ...  The decay width difference is found to be Delta Gamma_s = 0.075 +/- 0.035 (stat) +/- 0.01 syst 1/ps.  ...  Kinematic, topological and particle identification information (time of flight and energy loss) are combined into an artificial neural network optimized for B s → J/ψφ candidate selection.  ... 
arXiv:1012.0962v2 fatcat:zvj2jpzn3jdipfyq3okre2m3se

High-speed wide-field time-gated endoscopic fluorescence-lifetime imaging

J. Requejo-Isidro, J. McGinty, I. Munro, D. S. Elson, N. P. Galletly, M. J. Lever, M. A. A. Neil, G. W. H. Stamp, P. M. W. French, P. A. Kellett, J. D. Hares, A. K. L. Dymoke-Bradshaw
2004 Optics Letters  
We thank Smith & Nephew plc for donating the endoscope. J. Requejo-Isidro's e-mail address is j.requejo@ imperial.ac.uk.  ...  The system is based on a GOI with a rapidly adjustable electronic delay circuit to permit optimization of the lifetime contrast for different f luorophores.  ...  or a high-frequency modulated source. 4 Alternatively, one can determine lifetime in the time domain by directly measuring the decay in f luorescence after a pulsed excitation, using time-gated imaging  ... 
doi:10.1364/ol.29.002249 pmid:15524370 fatcat:c676fonmrfacjfkuv4psqppgoi
« Previous Showing results 1 — 15 out of 66,528 results