• Electronics and Telecommunications Trends
• /
• v.36 no.2
• /
• pp.32-42
• /
• 2021

The rapid growth of deep-learning applications has invoked the R&D of artificial intelligence (AI) processors. A dedicated software framework such as a compiler and runtime APIs is required to achieve maximum processor performance. There are various compilers and frameworks for AI training and inference. In this study, we present the features and characteristics of AI compilers, training frameworks, and inference engines. In addition, we focus on the internals of compiler frameworks, which are based on either basic linear algebra subprograms or intermediate representation. For an in-depth insight, we present the compiler infrastructure, internal components, and operation flow of ETRI's "AI-Ware." The software framework's significant role is evidenced from the optimized neural processing unit code produced by the compiler after various optimization passes, such as scheduling, architecture-considering optimization, schedule selection, and power optimization. We conclude the study with thoughts about the future of state-of-the-art AI compilers.

• International Journal of Computer Science & Network Security
• /
• v.23 no.8
• /
• pp.85-90
• /
• 2023

The Ant Colony System (ACS) is a variant of Ant colony optimization algorithm which is well-known in Traveling Salesman Problem. This paper proposed a hybrid method based on genetic algorithm (GA) and ant colony system (ACS), called GACS, to solve traffic routing problem. In the GACS, we use genetic algorithm to optimize the ACS parameters that aims to attain the shortest trips and time through new functions to help the ants to update global and local pheromones. Our experiments are performed by the GACS framework which is developed from VANETsim with the ability of real map loading from open street map project, and updating traffic light in real-time. The obtained results show that our framework acquired higher performance than A-Star and classical ACS algorithms in terms of length of the best global tour and the time for trip.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.9
• /
• pp.339-348
• /
• 2021

In recent software defect prediction research, defect prediction between cross projects and cross-version projects are actively studied. Cross-version defect prediction studies assume WP(Within-Project) so far. However, in the CV(Cross-Version) environment, the previous work does not consider the distribution difference between project versions is important. In this study, we propose an automated Bayesian optimization framework that considers distribution differences between different versions. Through this, it automatically selects whether to perform transfer learning according to the difference in distribution. This framework is a technique that optimizes the distribution difference between versions, transfer learning, and hyper-parameters of the classifier. We confirmed that the method of automatically selecting whether to perform transfer learning based on the distribution difference is effective through experiments. Moreover, we can see that using our optimization framework is effective in improving performance and, as a result, can reduce software inspection effort. This is expected to support practical quality assurance activities for new version projects in a cross-version project environment.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.24 no.1
• /
• pp.46-51
• /
• 2014

The most important process of engineering system optimal design is to identify the relationship between the design variables and system response. In case of the system optimization, Response Surface Method (RSM) is widely used. The optimization process of RSM generates the design space using the typical alternative candidates and finds the optimal design point in the generated design space. By changing the optimal point depending on the configuration of the design space, it is important to generate the design space. Therefor in this study, the design space is generated by using the relationship between design variables and system response based on Neuro-Response Surface Method (NRSM). And I try to construct the framework for optimal shape design based on NRSM that the optimum shape can be predicted using the Non-dominated Sorting Genetic Algorithm-II (NSGA-II) within the generated design space. In order to verify the usefulness of the constructed framework, we applied the nonlinear mathematical function problem. In this study, we can solve the constraints of time in the optimization process for the engineering problem and effective to determine the optimal design was possible. by using the generated framework for optimal shape design based on NRSM. In the future research, we try to apply the optimization problem for Naval Architectural & Ocean Engineering based on the results of this study.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.32 no.10
• /
• pp.12-19
• /
• 2004

An aerodynamic design optimization system for three-dimensional wing was developed as a part of the future MDO framework. The present design optimization system includes four modules such as geometry design, grid generation, flow solver and optimizer. All modules were based on commercial softwares and programmed to have automated execution capability in batch mode utilizing built-in script and journaling. The integration of all modules into the system was accomplished through programming using Visual Basic language. The distributed computational environment based on network communication was established to save computational time especially for time-consuming aerodynamic analyses. The distributed aerodynamic computations were performed in conjunction with the global optimization algorithm of response surface method, instead of using usual parallel computation based on domain decomposition. The application of the design system in the drag minimization problem demonstrated considerably enhanced efficiency of the design process while the final design showed reasonable results of reduced drag.

• Geomechanics and Engineering
• /
• v.34 no.6
• /
• pp.597-609
• /
• 2023

One of the applicable methods for the stabilization of soil walls is the nailing system which consists of tensile struts. The stability and safety of soil nail wall systems are influenced by the geometrical parameters of the nailing system. Generally, the determination of nailing parameters in order to achieve optimal performance of the nailing system for the safety of soil walls is defined in the framework of optimization problems. Also, according to the various uncertainty in the mechanical parameters of soil structures, it is necessary to evaluate the reliability of the system as a probabilistic problem. In this paper, the optimal design of the nailing system is carried out in deterministic and probabilistic cases using meta-heuristic and reliability-based design optimization methods. The colliding body optimization algorithm and first-order reliability method are used for optimization and reliability analysis problems, respectively. The objective function is defined based on the total cost of nails and safety factors and reliability index are selected as constraints. The mechanical properties of the nailing system are selected as design variables and the mechanical properties of the soil are selected as random variables. The results show that the reliability of the optimally designed soil nail system is very sensitive to uncertainty in soil mechanical parameters. Also, the design results are affected by uncertainties in soil mechanical parameters due to the values of safety factors. Reliability-based design optimization results show that a nailing system can be designed for the expected level of reliability and failure probability.

• Smart Structures and Systems
• /
• v.33 no.3
• /
• pp.177-188
• /
• 2024

The fatigue-induced sequential failure of a structure having structural redundancy requires system-level analysis to account for stress redistribution. System reliability-based design optimization (SRBDO) for preventing fatigue-initiated structural failure is numerically costly owing to the inclusion of probabilistic constraints. This study incorporates the Branch-and-Bound method employing system reliability Bounds (termed the B³ method), a failure-path structural system reliability analysis approach, with a metaheuristic optimization algorithm, namely grey wolf optimization (GWO), to obtain the optimal design of structures under fatigue-induced system failure. To further improve the efficiency of this new optimization framework, an additional bounding rule is proposed in the context of SRBDO against fatigue using the B³ method. To demonstrate the proposed method, it is applied to complex problems, a multilayer Daniels system and a three-dimensional tripod jacket structure. The system failure probability of the optimal design is confirmed to be below the target threshold and verified using Monte Carlo simulation. At earlier stages of the optimization, a smaller number of limit-state function evaluation is required, which increases the efficiency. In addition, the proposed method can allocate limited materials throughout the structure optimally so that the optimally-designed structure has a relatively large number of failure paths with similar failure probability.

• International Journal of Fluid Machinery and Systems
• /
• v.9 no.2
• /
• pp.150-158
• /
• 2016

In this paper, a new framework that works the automatic execution with less design cycle time and human intervention bottlenecks is introduced to optimize the vortex shedder design by numerical integration method. This framework is based on iSIGHT combined with the pre-processor GAMBIT, and flow analysis software FLUENT. Two vortex shedders, circular with slit and triangular- semi circular cylinder, are employed as the designed models to be optimized, and DOE driver is used for optimization. According to the essential properties of a vortex shedder, it has found that the best diameters are 30mm for circular cylinder with slit and 30 to 35 mm for tri-semi cylinder. For slit ratio, 0.1 and 0.15 are the optimized values for circular with slit and tri-semi cylinder respectively. And it is found that these optimal results generated by DOE automated design cycle are in well agreement with the experiment.

• Earthquakes and Structures
• /
• v.11 no.4
• /
• pp.701-721
• /
• 2016

Base isolation is a well-established passive strategy for seismic response control of buildings. In this paper, an efficient framework is proposed for reliability-based design optimization (RBDO) of isolated buildings subjected to uncertain earthquakes. The framework uses reduced function evaluations method, as an efficient tool for structural reliability analysis, and an efficient optimization algorithm for optimal structural design. The probability of failure is calculated considering excessive base displacement, superstructure inter-storey drifts, member stress ratios and absolute accelerations of floors of the isolated building as failure events. The behavior of rubber bearing isolators is modeled using nonlinear hysteretic model and the variability of future earthquakes is modeled by applying a probabilistic approach. The effects of pulse component of stochastic near-fault ground motions, fixity-factor of semi-rigid beam-to-column connections, values of isolator parameters, earthquake magnitude and epicentral distance on the performance and safety of semi-rigidly connected base-isolated steel framed buildings are studied. Suitable RBDO examples are solved to illustrate the results of investigations.

• ETRI Journal
• /
• v.44 no.5
• /
• pp.746-758
• /
• 2022

This paper focuses on a mobile edge-computing-enabled heterogeneous network. A battery level-aware task-scheduling framework is proposed to improve the energy efficiency and prolong the operating hours of battery-powered mobile devices. The formulated optimization problem is a typical mixed-integer nonlinear programming problem. To solve this nondeterministic polynomial (NP)-hard problem, a decomposition-based task-scheduling algorithm is proposed. Using an alternating optimization technology, the original problem is divided into three subproblems. In the outer loop, task offloading decisions are yielded using a pruning search algorithm for the task offloading subproblem. In the inner loop, closed-form solutions for computational resource allocation subproblems are derived using the Lagrangian multiplier method. Then, it is proven that the transmitted power-allocation subproblem is a unimodal problem; this subproblem is solved using a gradient-based bisection search algorithm. The simulation results demonstrate that the proposed framework achieves better energy efficiency than other frameworks. Additionally, the impact of the battery level-aware scheme on the operating hours of battery-powered mobile devices is also investigated.