人工智慧實作教學研究：以電腦視覺為例

Abstract

人工智慧是未來科技發展的主軸，因此，各國皆大力進行人工智慧人才培養，國內大專院校各科系亦大量開課以因應產業需求，特別是列為必修課程，然而，人工智慧本身具有深奧的學理內涵，學生學習此項技術相對困難許多，教學現場急需尋求恰當的教學方法以提升學生的學習成效。有鑑於此，本研究採用問題引導教學策略並發展其教材，以實現電腦視覺為主題，過程中輔以問題引導，適度解決學生背景知識不足，進而提升學習成效。本研究之問題引導包括：如何將數學轉化成程式碼進行計算、人工智慧演算法程式寫法以及人工智慧實作，為利於學生學習，本研究使用Google Colab網頁互動程式進行教學，讓學習者進行機器學習的訓練集、測試集與深度學習。關於人工智慧教學實作的學習成效，採用質性方式進行，透過訪談學生使用問題引導教學策略後，學生反饋的資料顯示，本研究所提出的方法對於學習程式設計有正向的幫助，而且有助於：思考邏輯進步、提高尋找答案的能力、想法更加多元化、團隊的溝通能力，而且能與他人互相討論，找出彼此想法或觀念知識不足的地方，達到自我反思的效果。關鍵字：問題引導教學策略、人工智慧教學、學習成效

Artificial intelligence is the main axis of future technological development. Therefore, all countries are vigorously cultivating artificial intelligence talents. Various departments in domestic colleges and universities also offer a large number of courses to meet the needs of the industry, especially as compulsory courses. However, artificial intelligence itself has depth in core beliefs and the base of knowledge. It is relatively difficult for students to learn this technology because of the theoretical connotation, and there is an urgent need to find appropriate teaching methods to improve student's learning effectiveness. In view of this, this study adopts the problem-based learning strategy and develops its teaching materials, with the theme of realizing computer vision, supplemented by problem-guided in the process, to moderately solve the lack of students' background knowledge, and then improve the learning effect. The problem-based learning includes: how to convert mathematics into code for calculation, artificial intelligence algorithm programming and artificial intelligence implementation. To facilitate students' learning, this study uses Google Colab web page interactive program for teaching, allowing learners to perform machine Learning training set, test set and deep learning. Regarding the learning effect of artificial intelligence teaching practice, a qualitative method was used. After interviewing students and using questions to guide teaching strategies, the feedback from students showed that the method proposed in this study is of positive help to learning programming, and has contribute to: thinking logical progress, improving the ability to find answers, more diverse ideas, team communication skills, and being able to discuss with others, find out where each other's ideas or conceptual knowledge is insufficient, and achieve the effect of self-reflection.

Keywords: problem-based learning strategies, artificial intelligence teaching, learning effectiveness