Reinforcement Learning Based Quantization Strategy Optimal Assignment Algorithm for Mixed Precision

The quantization algorithm compresses the original network by reducing the numerical bit width of the model,which improves the computation speed. Because different layers have different redundancy and sensitivity to databit width. Reducing the data bit width will result in a loss of accuracy. Therefore, it is difficult to determinethe optimal bit width for different parts of the network with guaranteed accuracy. Mixed precision quantizationcan effectively reduce the amount of computation while keeping the model accuracy basically unchanged. In thispaper, a hardware-aware mixed precision quantization strategy optimal assignment algorithm adapted to low bitwidth is proposed, and reinforcement learning is used to automatically predict the mixed precision that meets theconstraints of hardware resources. In the state-space design, the standard deviation of weights is used to measurethe distribution difference of data, the execution speed feedback of simulated neural network accelerator inferenceis used as the environment to limit the action space of the agent, and the accuracy of the quantization model afterretraining is used as the reward function to guide the agent to carry out deep reinforcement learning training. Theexperimental results show that the proposed method obtains a suitable model layer-by-layer quantization strategyunder the condition that the computational resources are satisfied, and themodel accuracy is effectively improved.The proposed method has strong intelligence and certain universality and has strong application potential in thefield of mixed precision quantization and embedded neural network model deployment.

Competition assignment problem algorithm based on Hungarian method

Traditional Hungarian method can only solve standard assignment problems, while can not solve competition assignment problems. This article emphatically discussed the difference between standard assignment problems and competition assignment problems. The kinds of competition assignment problem algorithms based on Hungarian method and the solutions of them were studied.

Storage Assignment Optimization in a Multi-tier Shuttle Warehousing System

The current mathematical models for the storage assignment problem are generally established based on the traveling salesman problem（TSP）,which has been widely applied in the conventional automated storage and retrieval system（AS/RS）.However,the previous mathematical models in conventional AS/RS do not match multi-tier shuttle warehousing systems（MSWS） because the characteristics of parallel retrieval in multiple tiers and progressive vertical movement destroy the foundation of TSP.In this study,a two-stage open queuing network model in which shuttles and a lift are regarded as servers at different stages is proposed to analyze system performance in the terms of shuttle waiting period（SWP） and lift idle period（LIP） during transaction cycle time.A mean arrival time difference matrix for pairwise stock keeping units（SKUs） is presented to determine the mean waiting time and queue length to optimize the storage assignment problem on the basis of SKU correlation.The decomposition method is applied to analyze the interactions among outbound task time,SWP,and LIP.The ant colony clustering algorithm is designed to determine storage partitions using clustering items.In addition,goods are assigned for storage according to the rearranging permutation and the combination of storage partitions in a 2D plane.This combination is derived based on the analysis results of the queuing network model and on three basic principles.The storage assignment method and its entire optimization algorithm method as applied in a MSWS are verified through a practical engineering project conducted in the tobacco industry.The applying results show that the total SWP and LIP can be reduced effectively to improve the utilization rates of all devices and to increase the throughput of the distribution center.

Improved Hungarian algorithm for assignment problems of serial-parallel systems

In order to overcome the shortcoming of the classical Hungarian algorithm that it can only solve the problems where the total cost is the sum of that of each job, an improved Hungarian algorithm is proposed and used to solve the assignment problem of serial-parallel systems. First of all, by replacing parallel jobs with virtual jobs, the proposed algorithm converts the serial-parallel system into a pure serial system, where the classical Hungarian algorithm can be used to generate a temporal assignment plan via optimization. Afterwards, the assignment plan is validated by checking whether the virtual jobs can be realized by real jobs through local searching. If the assignment plan is not valid, the converted system will be adapted by adjusting the parameters of virtual jobs, and then be optimized again. Through iterative searching, the valid optimal assignment plan can eventually be obtained.To evaluate the proposed algorithm, the valid optimal assignment plan is applied to labor allocation of a manufacturing system which is a typical serial-parallel system.

New optimal algorithm of data association for multi-passive-sensor location system

In dense target and false detection scenario of four time difference of arrival （TDOA） for multi-passive-sensor location system, the global optimal data association algorithm has to be adopted. In view of the heavy calculation burden of the traditional optimal assignment algorithm, this paper proposes a new global optimal assignment algorithm and a 2-stage association algorithm based on a statistic test. Compared with the traditional optimal algorithm, the new optimal algorithm avoids the complicated operations for finding the target position before we calculate association cost; hence, much of the procedure time is saved. In the 2-stage association algorithm, a large number of false location points are eliminated from candidate associations in advance. Therefore, the operation is further decreased, and the correct data association probability is improved in varying degrees. Both the complexity analyses and simulation results can verify the effectiveness of the new algorithms.

AN ANT COLONY ALGORITHM FOR YARD TRUCK SCHEDULING AND YARD LOCATION ASSIGNMENT PROBLEMS WITH PRECEDENCE CONSTRAINTS

This paper examines the yard truck scheduling,the yard location assignment for discharging containers,and the quay crane scheduling in container terminals.Taking into account the practical situation,we paid special attention to the loading and discharging precedence relationships between containers in the quay crane operations.A Mixed Integer Program（MIP） model is constructed,and a two-stage heuristic algorithm is proposed.In the first stage an Ant Colony Optimization（ACO） algorithm is employed to generate the yard location assignment for discharging containers.In the second stage,the integration of the yard truck scheduling and the quay crane scheduling is a flexible job shop problem,and an efficient greedy algorithm and a local search algorithm are proposed. Extensive numerical experiments are conducted to test the performance of the proposed algorithms.

Chanel Assignment Optimization Algorithm Based on Link Throughput and Node Priority

How to reduce interference among neighbor nodes in wireless mesh networks is still an important and key issue nowa- days. In this paper, an optimized channel assignment algorithm （OCA） is proposed to solve this problem based on link throughput and node priority. The effects of the numbers of network interface cards and channels on the network throughput are analyzed and evaluated, When there are seven of the numbers of both network interface cards and channels, the efficiency of utilizing network interface card and channel reaches highest. Compared with cen- tralized channel assignment algorithm （CCA）, the proposed algo- rithm has less packet loss rate and more network throughput sig- nificantly.

Web-based construction equipment fleet management system: cost-effective global and local allocation

Over the last two decades, construction contractors have been gradually making more investments in construction equipment to meet their needs associated with increasing volumes of construction projects. At present,from an operational perspective, almost all contractors pay more attention to maintaining their equipment fleets in well-sustained workable conditions and having a high accessibility of the necessary equipment pieces. However,such an approach alone is not enough to maintain an efficient and sustainable business. In particular, for largescale construction companies that operate in multiple sites in the U.S. or overseas, the problem extends to an optimal allocation of available equipment. Given the current state of the construction industry in the U.S., this problem can be solved by geographically locating equipment pieces and then wisely re-allocating them among projects. Identifying equipment pieces geographically is a relatively easy task.The difficulty arises when informed decision-making is required for equipment allocation among job sites. The actual allocation of equipment should be both economically feasible and technologically preferable. To help in informed decision-making, an optimization model is developed as a mixed integer program. This model is formed based on a previously successfully developed decision-support model for construction equipment selection. The proposed model incorporates logical strategies of supply chain management to optimally select construction equipment for any construction site while taking into account the costs, availability, and transportation-relatedissues as constraints. The model benefits those responsible for informed decision-making for construction equipment selection and allocation. It also benefits the owners of construction companies, owing to its cost-minimization objective.