A Forward-Propagation Rule for Acquiring Neural Inverse Models Using a RLS Algorithm

2004 ◽  
pp. 585-591 ◽  
Author(s):  
Yoshihiro Ohama ◽  
Naohiro Fukumura ◽  
Yoji Uno
Keyword(s):  
Rls Algorithm ◽  
Inverse Models ◽  
Propagation Rule

Curve Estimation Modeling for Predictions of the Novel Coronavirus (Covid-19) Epidemic in Nigeria

2020 ◽  
Vol 4 (1) ◽  
pp. 1
Author(s):  
Christian Ebere Enyoh ◽  
Andrew Wirnkor Verla ◽  
Chidi Edbert Duru ◽  
Emmanuel Chinedu Enyoh ◽  
Budi Setiawan
Keyword(s):  
State Governments ◽  
The Novel ◽  
Curve Estimation ◽  
Exponential Models ◽  
Inverse Models ◽  
Forward Prediction ◽  
S Curve ◽  
Novel Coronavirus ◽  
Best Fit ◽  
Official Data

Based on the official Nigeria Centre for Disease Control (NCDC) data, the current research paper modeled the confirmed cases of the novel coronavirus disease 2019 (COVID-19) in Nigeria. Ten different curve regression models including linear, logarithmic, inverse, quadratic, cubic, compound, power, S-curve, growth, and exponential were used to fit the obtained official data. The cubic (R2 = 0.999) model gave the best fit for the entire country. However, the growth and exponential had the lowest standard error of estimate (0.958) and thus may best be used. The equations for these models were e0.78897+0.0944x and 2.2011e0.0944x respectively. In terms of confirmed cases in individual State, quadratic, cubic, compound, growth, power and exponential models generally best describe the official data for many states except for the state of Kogi which is best fitted with S-curve and inverse models.  The error between the model and the official data curve is quite small especially for compound, power, growth and exponential models. The computed models will help to realized forward prediction and backward inference of the epidemic situation in Nigeria, and the relevant analysis help Federal and State governments to make vital decisions on how to manage the lockdown in the country.

scholarly journals Personalized project recommendations: using reinforcement learning

2019 ◽  
Vol 2019 (1) ◽  
Author(s):  
Faxin Qi ◽  
Xiangrong Tong ◽  
Lei Yu ◽  
Yingjie Wang
Keyword(s):  
Reinforcement Learning ◽  
User Behavior ◽  
Collaborative Work ◽  
Recursive Least Squares ◽  
The Internet ◽  
Dynamic Impact ◽  
Rls Algorithm ◽  
Trust Value ◽  
Q Learning ◽  
Actual Evaluation

AbstractWith the development of the Internet and the progress of human-centered computing (HCC), the mode of man-machine collaborative work has become more and more popular. Valuable information in the Internet, such as user behavior and social labels, is often provided by users. A recommendation based on trust is an important human-computer interaction recommendation application in a social network. However, previous studies generally assume that the trust value between users is static, unable to respond to the dynamic changes of user trust and preferences in a timely manner. In fact, after receiving the recommendation, there is a difference between actual evaluation and expected evaluation which is correlated with trust value. Based on the dynamics of trust and the changing process of trust between users, this paper proposes a trust boost method through reinforcement learning. Recursive least squares (RLS) algorithm is used to learn the dynamic impact of evaluation difference on user’s trust. In addition, a reinforcement learning method Deep Q-Learning (DQN) is studied to simulate the process of learning user’s preferences and boosting trust value. Experiments indicate that our method applied to recommendation systems could respond to the changes quickly on user’s preferences. Compared with other methods, our method has better accuracy on recommendation.

Note on Niederreiter-Xing's Propagation Rule for Linear Codes

2002 ◽  
Vol 13 (1) ◽  
pp. 53-56 ◽  
Author(s):  
Ferruh Özbudak ◽  
Henning Stichtenoth
Keyword(s):  
Linear Codes ◽  
Propagation Rule

scholarly journals Can We Infer Diapycnal Mixing Rates from the World Ocean Temperature–Salinity Distribution?

2016 ◽  
Vol 46 (12) ◽  
pp. 3751-3775 ◽  
Author(s):  
Olivier Arzel ◽  
Alain Colin de Verdière
Keyword(s):  
General Circulation ◽  
Large Scale ◽  
Water Mass ◽  
World Ocean ◽  
Eddy Diffusivity ◽  
Regional Variability ◽  
Diapycnal Mixing ◽  
Inverse Models ◽  
The World ◽  
The Difference

AbstractThe turbulent diapycnal mixing in the ocean is currently obtained from microstructure and finestructure measurements, dye experiments, and inverse models. This study presents a new method that infers the diapycnal mixing from low-resolution numerical calculations of the World Ocean whose temperatures and salinities are restored to the climatology. At the difference of robust general circulation ocean models, diapycnal diffusion is not prescribed but inferred. At steady state the buoyancy equation shows an equilibrium between the large-scale diapycnal advection and the restoring terms that take the place of the divergence of eddy buoyancy fluxes. The geography of the diapycnal flow reveals a strong regional variability of water mass transformations. Positive values of the diapycnal flow indicate an erosion of a deep-water mass and negative values indicate a creation. When the diapycnal flow is upward, a diffusion law can be fitted in the vertical and the diapycnal eddy diffusivity is obtained throughout the water column. The basin averages of diapycnal diffusivities are small in the first 1500 m [O(10−5) m2 s−1] and increase downward with bottom values of about 2.5 × 10−4 m2 s−1 in all ocean basins, with the exception of the Southern Ocean (50°–30°S), where they reach 12 × 10−4 m2 s−1. This study confirms the small diffusivity in the thermocline and the robustness of the higher canonical Munk’s value in the abyssal ocean. It indicates that the upward dianeutral transport in the Atlantic mostly takes place in the abyss and the upper ocean, supporting the quasi-adiabatic character of the middepth overturning.

Comparative analysis of power amplifiers' polynomial based models identification using RLS algorithm

2016 ◽  
Author(s):  
Abubaker Abdelhafiz ◽  
Fadhel M. Ghannouchi ◽  
Oualid Hammi ◽  
Azzedine Zerguine
Keyword(s):  
Comparative Analysis ◽  
Power Amplifiers ◽  
Rls Algorithm

On the effect of input signal correlation on weight misadjustment in the RLS algorithm

1995 ◽  
Vol 43 (4) ◽  
pp. 988-991 ◽  
Author(s):  
T. Adali ◽  
S.H. Ardalan
Keyword(s):  
Input Signal ◽  
Rls Algorithm ◽  
Signal Correlation

Some Tricks of Calculations in MIL RLS Algorithm

2021 ◽  
Author(s):  
Victor Djigan
Keyword(s):  
Rls Algorithm

A Comparison of Inverse Models for the Estimation of Ice-Induced Propeller Moments on a Polar Vessel

2021 ◽  
Author(s):  
Brendon M. Nickerson ◽  
Anriëtte Bekker
Keyword(s):  
Rotational Speed ◽  
Continuous Model ◽  
Full Scale ◽  
Lumped Mass ◽  
Mass Model ◽  
Inverse Models ◽  
Lumped Mass Model ◽  
Ill Posed ◽  
Shaft Torque ◽  
Measurement Location

Abstract Full-scale measurements were conducted on the port side propulsion shaft the S.A. Agulhas II during the 2019 SCALE Spring Cruise. The measurements included the shaft torque captured at two separate measurement locations, and the shaft rotational speed at one measurement location. The ice-induced propeller moments are estimated from the full-scale shaft responses using two inverse models. The first is a published discrete lumped mass model that relies on regularization due to the inverse problem being ill-posed. This model is only able to make use of the propulsion shaft torque as inputs. The second model is new and employs modal superposition to represent the propulsion shaft as a combination of continuous modes, resulting in a well-posed problem. This new model requires the additional measurement of the shaft rotational speed for the inverse solution. The continuous model is shown to be more consistent and efficient, which allows its use in real-time monitoring of propeller moments.

Tectonic controls on basement exhumation in the southern Rocky Mountains (United States): The power of combined zircon (U-Th)/He and K-feldspar 40Ar/39Ar thermochronology

Geology ◽  
2021 ◽  
Author(s):  
Jason W. Ricketts ◽  
Jacoup Roiz ◽  
Karl E. Karlstrom ◽  
Matthew T. Heizler ◽  
William R. Guenthner ◽  
...  
Keyword(s):  
Rocky Mountains ◽  
Rocky Mountain ◽  
Snowball Earth ◽  
Southern Rocky Mountains ◽  
Basement Rocks ◽  
Rocky Mountain Region ◽  
Multi Stage ◽  
Inverse Models ◽  
Multiple Pulses ◽  
Differential Uplift

The Great Unconformity of the Rocky Mountain region (western North America), where Precambrian crystalline basement is nonconformably overlain by Phanerozoic strata, represents the removal of as much as 1.5 b.y. of rock record during 10-km-scale basement exhumation. We evaluate the timing of exhumation of basement rocks at five locations by combining geologic data with multiple thermochronometers. 40Ar/39Ar K-feldspar multi-diffusion domain (MDD) modeling indicates regional multi-stage basement cooling from 275 to 150 °C occurred at 1250–1100 Ma and/or 1000–700 Ma. Zircon (U-Th)/He (ZHe) dates from the Rocky Mountains range from 20 to 864 Ma, and independent forward modeling of ZHe data is also most consistent with multi-stage cooling. ZHe inverse models at five locations, combined with K-feldspar MDD and sample-specific geochronologic and/or thermochronologic constraints, document multiple pulses of basement cooling from 250 °C to surface temperatures with a major regional basement exhumation event 1300–900 Ma, limited cooling in some samples during the 770–570 Ma breakup of Rodinia and/or the 717–635 Ma snowball Earth, and ca. 300 Ma Ancestral Rocky Mountains cooling. These data argue for a tectonic control on basement exhumation leading up to formation of the Precambrian-Cambrian Great Unconformity and document the formation of composite erosional surfaces developed by faulting and differential uplift.

Sign in / Sign up

Export Citation Format

Share Document