Gradient Reinforcement Research Articles

A nonlinear total variation based computed tomography (CT) image reconstruction method using gradient reinforcement.

Compressed sensing-based reconstruction algorithms have been proven to be more successful than analytical or iterative methods for sparse computed tomography (CT) imaging by narrowing down the solution set thanks to its ability to seek a sparser solution. Total variation (TV), one of the most popular sparsifiers, exploits spatial continuity of features by restricting variation between two neighboring pixels in each direction as using partial derivatives. When the number of projections is much fewer than the one in conventional CT, which results in much less sampling rate than the minimum required one, TV may not provide satisfactory results. In this study, a new regularizer is proposed which seeks for a sparser solution by reinforcing the gradient of TV and empowering the spatial continuity of features. The experiments are done by using both analitical phantom and real human CT images and the results are compared with conventional, four-directional, and directional TV algorithms by using contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR) and Structural Similarity Index (SSIM) metrics. Both quantitative and visual evaluations show that the proposed method is promising for sparse CT image reconstruction by reducing the background noise while preserving the features and edges.

FULL-STACK розробка інтелектуальної системи розвитку міграції населення

A generalised model of population migration is proposed. On the basis of the model of population migration, the article develops models of: a set of directions of population flows, which are formed on internal and external flows of the State; duration of migration, which is determined by its nature in time, including permanent or irreversible duration of migration, movement for a relatively short time, annual movement of people and pendulum duration of migration; type and form of migration. A model of indicators of actual migration (resettlement) that can characterise the overall level of mobility of the population of the territories, the scale, structure, directions and effectiveness of migration flows for a given period is developed and their groups are divided. It is proposed that the results of population migration should be presented in the form of a number of absolute and relative indicators for the purpose of further regression analysis of data, namely, those who arrived for permanent residence from other settlements; those who left for permanent residence to other settlements; migration balance or mechanical growth. Inter-rayon relations are characterised by the strength of migration flows. To obtain the results of migration, we take into account the strength of migration flows, which depend on the population of the territories between which the exchange takes place and on their location. The result of this exchange is expressed in the migration balance or by means of efficiency coefficients of migration ties. The intensity of migration exchange, independent of the population size of both the areas of origin and the places of settlement, is determined by the intensity coefficients of migration ties. The types of migration intensity coefficients are formed depending on the properties, namely the intensity coefficients of arrival (immigration), departure (emigration), reverse migration, and net migration. The intelligent geographic information system implements the lightgbm algorithm for population migration forecasting, which is a decision tree with gradient reinforcement. For data analysis, the migration forecasting system includes regression analysis and neural networks and is capable of predicting international migration or migration between different countries.

Robot arm navigation using deep deterministic policy gradient algorithms

ABSTRACT In this paper, the Deep Deterministic Policy Gradient (DDPG) reinforcement learning algorithm is employed to enable a double-jointed robot arm to reach continuously changing target locations. The experimentation of the algorithm is carried out by training an agent to control the movement of this double-jointed robot arm. The architectures of the actor and cretic networks are meticulously designed and the DDPG hyperparameters are carefully tuned. An enhanced version of the DDPG is also presented to handle multiple robot arms simultaneously. The trained agents are successfully tested in the Unity Machine Learning Agents environment for controlling both a single robot arm as well as multiple simultaneous robot arms. The testing shows the robust performance of the DDPG algorithm for empowering robot arm manoeuvring in complex environments.

Étude expérimentale sur les poutres mixtes

This paper presents a summary of the first part of a general research program on the nonlinear theory of composite beams and columns. The paper deals with the experimental study of simply supported composite beams, submitted to uniaxial bending about the major axis. Test results are compared with theoretical results obtained from an elaborate nonlinear analysis.A total of eight composite beams have been tested, four of them with special dowels welded to the wide-flange section web, and steel plates welded at both ends, in order to limit sliding between steel and concrete.By adequately adjusting the concrete stress–strain curve measured on control cylinders, to take into account the effects of strain rate, strain gradient, and steel reinforcement confinement, we obtain a theoretical ultimate bending moment identical to the experimental moment, and a deflection at rupture 10% smaller than the measured one. Key words: reinforced concrete, stress–strain curve, bending, ultimate moment, ultimate deflection, nonlinear theory, composite beams, composite sections.