Search

Summary: As a deep neural networks (DNNs) model compression method, learning‐compression (LC) algorithm based on pre‐trained models and matrix decomposition increases training time and ignores the structural information of models. In this manuscript, a tensor decomposition‐based direct LC (TDLC) algorithm without pre‐trained models is proposed. In TDLC, the pre‐trained model is eliminated, and tensor decomposition is first applied to LC algorithm to preserve the structural features of the model. There are two key steps in TDLC. An optimal rank selection method is first proposed in compression‐step (C‐step) of TDLC to find global optimal ranks of tensor decomposition. Second, TDLC utilizes cyclical learning rate, which is different from traditional monotonically learning rates schedule, to improve the generalization performance of uncompressed models in learning‐step (L‐step). TDLC obtains the optimal compression model by alternately optimizing L‐step and C‐step. TDLC is compared with 16 state‐of‐the‐art compression methods in experiments part. Extensive experimental results show that TDLC produces high‐accuracy compression models with high compression rate. Comparing with TDLC‐pre‐trained, TDLC notably achieves 30% training time shorten and 11% parameter reduction in Resnet32, while improving accuracy by 0.2%. [ABSTRACT FROM AUTHOR]

The ability of a species to colonize newly available habitats is crucial to its overall fitness.sup.1-3. In general, motility and fast expansion are expected to be beneficial for colonization and hence for the fitness of an organism.sup.4-7. Here we apply an evolution protocol to investigate phenotypical requirements for colonizing habitats of different sizes during range expansion by chemotaxing bacteria.sup.8. Contrary to the intuitive expectation that faster is better, we show that there is an optimal expansion speed for a given habitat size. Our analysis showed that this effect arises from interactions among pioneering cells at the front of the expanding population, and revealed a simple, evolutionarily stable strategy for colonizing a habitat of a specific size: to expand at a speed given by the product of the growth rate and the habitat size. These results illustrate stability-to-invasion as a powerful principle for the selection of phenotypes in complex ecological processes. To successfully colonize a habitat in the presence of competing chemotactic bacterial populations, the winner is required to expand its range at a speed given by the habitat size and the population doubling time., Author(s): Weirong Liu [sup.1] [sup.2] , Jonas Cremer [sup.3] [sup.4] , Dengjin Li [sup.1] [sup.2] , Terence Hwa [sup.3] , Chenli Liu [sup.1] [sup.2] Author Affiliations: (1) CAS Key Laboratory [...]

It is a challenge to maintain a safe and efficient tracking for multiple high-speed trains under the moving block operational mode. In this paper, a novel cooperative tracking control based on a consensus algorithm and artificial potential field theory is proposed to realize the train tracking within a distance range. A tunable artificial potential function is first designed to dynamically adjust the distance between adjacent high-speed trains with real-time train states. By regulating the parameters of the artificial potential function, the safety distance can be adjusted according to the required tolerance deviation of the actual distance. Under the proposed strategy, each high-speed train operates with the desired speed and tracks the preceding one with an adjustable distance range. Numerical train operational cases are investigated to illustrate the effectiveness of the proposed methods., Author(s): Zhiwu Huang [1]; Pingping Wang [1]; Feng Zhou (corresponding author) [2]; Weirong Liu [3]; Jun Peng [3] 1. Introduction Based on an advanced wireless communication technology and positioning technology, [...]

Currently, in most traditional VSLAM (visual SLAM) systems, static assumptions result in a low accuracy in dynamic environments, or result in a new and higher level of accuracy but at the cost of sacrificing the real–time property. In highly dynamic scenes, balancing a high accuracy and a low computational cost has become a pivotal requirement for VSLAM systems. This paper proposes a new VSLAM system, balancing the competitive demands between positioning accuracy and computational complexity and thereby further improving the overall system properties. From the perspective of accuracy, the system applies an improved lightweight target detection network to quickly detect dynamic feature points while extracting feature points at the front end of the system, and only feature points of static targets are applied for frame matching. Meanwhile, the attention mechanism is integrated into the target detection network to continuously and accurately capture dynamic factors to cope with more complex dynamic environments. From the perspective of computational expense, the lightweight network Ghostnet module is applied as the backbone network of the target detection network YOLOv5s, significantly reducing the number of model parameters and improving the overall inference speed of the algorithm. Experimental results on the TUM dynamic dataset indicate that in contrast with the ORB–SLAM3 system, the pose estimation accuracy of the system improved by 84.04%. In contrast with dynamic SLAM systems such as DS–SLAM and DVO SLAM, the system has a significantly improved positioning accuracy. In contrast with other VSLAM algorithms based on deep learning, the system has superior real–time properties while maintaining a similar accuracy index. [ABSTRACT FROM AUTHOR]

The traceability of blockchain is crucial to the quality and safety of agricultural products, primarily when bacterial contamination occurs in the agricultural supply chain. To gain a deeper understanding of the impacts on the quality and safety of agricultural products, we built a three-level agricultural supply chain model comprising one retailer, one manufacturer, and one supplier. We aimed to explore the impact of blockchain traceability and retailers' altruistic preference on pricing decisions and channel member profit, and to then examine the motivation of enterprises to participate in blockchain technology. Our study showed that: (1) blockchain technology has the potential to improve the prices and profits of supply chain members without considering the cost of blockchain; and (2) blockchain technology has the potential to improve the retailer's price when the unit variable cost of blockchain is higher than a certain threshold. It can also reduce the manufacturer's price and the supplier's output. However, the excessive cost of blockchain can also aggravate the double marginalization effect of the supply chain. When both the unit variable cost and the fixed input cost of the blockchain are low, each enterprise has the incentive to participate in the blockchain. (3) In the scenario of the altruistically-preferred retailer, when the blockchain variable cost is lower than a certain threshold, the retailer's altruistic preferences can improve the manufacturer's and supplier's profit, but it will reduce the retailer's profit. [ABSTRACT FROM AUTHOR]

Summary: This article deals with state estimation of complex nonlinear discrete fractional‐order systems with unknown noise statistics by means of an adaptive fractional‐order Unscented Kalman filter (AFUKF). Firstly, in order to alleviate the communication burden of fractional‐order Unscented Kalman filter, short‐term memory effect is utilized to decide an appropriate memory length. Then aiming at the problem of filtering divergence and accuracy degradation caused by unknown statistical characteristics of noise, based on the maximum a posterior (MAP) principle, a noise statistical estimator is introduced to estimate and correct the statistical characteristics of noise in real‐time. Finally, the unbiasedness of the proposed algorithm is analyzed to verify that the estimated mean and covariance of noise are unbiased. The effectiveness and accuracy of AFUKF are demonstrated via simulation experiments. [ABSTRACT FROM AUTHOR]

Supercapacitors undergo wide terminal voltage fluctuations, which makes it difficult to maintain a stable bus voltage and develop a satisfactory dynamic response. To solve this problem, a fast adaptive bus voltage regulation strategy is proposed in this paper. Firstly, the dual‐loop controller with an external voltage loop and an internal current loop is applied to the supercapacitor energy storage systems. Secondly, an adaptive proportional–integral controller is designed for the outer voltage loop to accommodate the variations of the system characteristics. Furthermore, in order to improve the dynamic response, the set‐point automatic modulation is integrated into the voltage control loop to compensate the stochastic load demanded fluctuations. The simulation and experimental results show that the proposed method ensures the stability of the bus voltage and provides a lower fluctuation and shorter settling time compared with the conventional method. [ABSTRACT FROM AUTHOR]

The high-speed train operation process is highly nonlinear and has multiple constraints and objectives, which lead to a requirement for the automatic train operation (ATO) system. In this paper, a hybrid model predictive control (MPC) framework is proposed for the controller design of the ATO system. Firstly, a piecewise linear system with state and input constraints is constructed through piecewise linearization of the high-speed train's nonlinear dynamics. Secondly, the piecewise linear system is transformed into a mixed logical dynamical (MLD) system by introducing the auxiliary binary variables. For the transformed MLD system, a hybrid MPC controller is designed to realize the precise control under hard constraints. To reduce the online computation complexity, the explicit control law is computed offline by employing the mixed-integer linear programming (MILP) technique. Simulation results validate the effectiveness of the proposed method., 1. Introduction With rapid development of high-speed railway, the operation safety, punctuality, and energy consumption of train have received more and more attentions. Nowadays, in China, the high-speed railway generally [...]

Aiming at the trajectory generation and optimization of mobile robots in complex and uneven environments, a hybrid scheme using mutual learning and adaptive ant colony optimization (MuL-ACO) is proposed in this paper. In order to describe the uneven environment with various obstacles, a 2D-H map is introduced in this paper. Then an adaptive ant colony algorithm based on simulated annealing (SA) is proposed to generate initial trajectories of mobile robots, where based on a de-temperature function of the simulated annealing algorithm, the pheromone volatilization factor is adaptively adjusted to accelerate the convergence of the algorithm. Moreover, the length factor, height factor, and smooth factor are considered in the comprehensive heuristic function of ACO to adapt to uneven environments. Finally, a mutual learning algorithm is designed to further smooth and shorten initial trajectories, in which different trajectory node sequences learn from each other to acquire the shortest trajectory sequence to optimize the trajectory. In order to verify the effectiveness of the proposed scheme, MuL-ACO is compared with several well-known and novel algorithms in terms of running time, trajectory length, height, and smoothness. The experimental results show that MuL-ACO can generate a collision-free trajectory with a high comprehensive quality in uneven environments. [ABSTRACT FROM AUTHOR]