標題: | 利用蟻群最適化協助員工選擇課程 Use Ant Colony Optimization to Assist Employees in Selecting the Training Courses |
作者: | 謝育勳 HSIEH YU HSUN 劉敦仁 林妙聰 LIU Duen Ren LIN BMT 管理學院資訊管理學程 |
關鍵字: | 螞蟻演算法;訓練課程規劃;決策支援;近似解法;Ant Colony Optimization;Training Course Planning;Decision Support;Approximation Algorithm |
公開日期: | 2005 |
摘要: | 在這資訊爆炸的時代裡,企業要如何維持其競爭力,不被市場所淘汰,這是現今企業最關心的議題。雖然不同的產業、市場定位,會有不同的競爭策略與方式,但所有的企業均會使其員工不斷的學習與成長,以因應不同的競爭環境。但在企業中,由於員工的時間與企業的預算是有限的,因此如何讓員工在有限的資源(時間與成本)下,參與最有價值的課程,是現今企業與員工最關心的課題。
本論文的主題是利用螞蟻演算法(Ant Colony Algorithm),協助員工做最適化的課程選擇。在所規劃之員工的課程集合中,必須考量企業所規定的訓練時間及預算下;另外,為了使員工不要只集中在某一個或兩個學習領域,因此也規定其所建議的訓練課程必須要涵蓋超過三個領域。本論文根據這些限制式及訓練課程資料,利用螞蟻演算法來產生最適化的課程建議,即求取『課程效用/(成本*時間)』之最大目標函數值。本論文亦將利用窮舉演算法(Exhaustive Algorithm),以驗證本論文所提演算法的效能(Effectiveness)跟效率(Efficiency)。在效能方面,由於窮舉法可提供最佳解,因此其可以精準驗證近似演算法的效度。在效率方面,本演算法屬於近似法,因此所花費的時間將遠低於窮舉法所使用的時間。
本研究還是有很多可以改進的空間,例如:介面的設計、管理功能的加強、與其他訓練系統(訓練系統或線上訓練系統)的結合以及驗證效率演算法的選擇(例如選擇更有效率的GA演算法),這些都是可以提供給未來研究者或是企業運用改良的方向。 In the information explosion era, the most critical issue for a company is how to keep its core competence and to prevent from losing the market shares. Companies in different industries and markets may deploy different strategies and methods to reinforce their competitiveness. Indeed, all companies will encourage their employees to continuously acquire new knowledge in the rapidly changing environment. Because there exist inevitable limitations in time and budget for preparing and selecting training courses, it is crucial to deploy some decision-aid mechanism to assist the employees to plan their training packages subject to the limited resources. In this thesis, we use Ant Colony Optimization (ACO) to provide advice for employees. The courses planning process needs to abide the company’s policies regarding training hours and costs. In order to avoid the courses of an employee are confined to only one or two areas, we stipulate that the selected training courses cover three or more areas. This study use ACO algorithm to generate packages of training courses for individual employees with the maximization of utility/(costs * hours) as the objective function values. Moreover, we also use an enumerate algorithm to verify the effectiveness and efficiency of our ACO algorithm. The two measurements are related to the solution quality and run time required by our algorithm. There are a lot of parts that it can be improved in the future work, for example: design the fancy user interface, enhance the management functions, link the other training system or E-learning system, and select the verification algorithm that it can exactly prove the system efficiency. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT009364508 http://hdl.handle.net/11536/79994 |
Appears in Collections: | Thesis |
Files in This Item:
If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.