MS004111-機器學習與進化計算

研究生課程開設申請表

 開課院（系、所）：beat365正版唯一       

 課程申請開設類型： 新開□     重開√     更名□（請在□内打勾，下同）

一、課程介紹（含教學目标、教學要求等）（300字以内）

本課程教學的目标是使學生掌握多種機器學習範型、算法及進化計算在機器學習領域的研究和應用，吸取包括概念學習、決策樹學習、人工神經網絡知識、統計和估計理論、貝葉斯觀點、計算學習理論、基于實例的學習方法、進化計算理論、學習規則集合的算法、分析學習、歸納與分析學習相結合以及增強學習方面的研究成果。要求通過本課程的教學，明确機器學習系統的幾個重要環節：選擇訓練經驗、目标函數、目标函數的表示、函數逼近算法，從而提高學生設計學習系統的能力， 增強學生對于機器學習這個多學科領域分析問題和解決問題的能力。

二、教學大綱（含章節目錄）：（可附頁）

（一）機器學習的概念（第一章）

包括學習問題的标準描述，設計學習系統的主要環節，機器學習中的經典問題和觀點。

（二）概念學習（第二章）

包括基于符号和邏輯表示的概念學習，假設的一般到特殊偏序結構和學習中引入歸納偏置的必要性。

（三）決策樹學習、人工神經網絡知識（第三章、第四章）

包括決策樹學習和過度拟合訓練數據的問題、人工神經網絡中的反向傳播算法以及梯度下降的一般方法。

（四）評估假設（第五章）

包括統計和估計理論的基礎概念，使用有限的樣本數據評估假設的精度。

（五）貝葉斯學習（第六章）

包括使用貝葉斯分析刻畫非貝葉斯學習算法以及直接處理概率的貝葉斯算法。

（六）計算學習理論（第七章）

包括可能近似正确學習模型和出錯界限學習模型及聯合多個學習方法的加權多數算法。

（七）基于實例的學習（第八章）

包括k-近鄰算法、局部加權回歸、徑向基函數以及基于案例的推理算法。

（八）進化計算（第九章）

包括進化計算的搜索策略和實現方法。

（九）學習規則集合（第十章）

包括序列覆蓋算法以及學習一階規則集方法理論。

（十）分析學習、歸納和分析學習的結合（第十一、十二章）

包括純粹的分析學習方法（基于解釋的學習）、結合分析和歸納學習以提高學習精度的方法。

（十一）增強學習（第十三章）

包括解決自治agent學習控制策略問題、馬爾可夫決策過程問題以及Q學習。

三、教學周曆

四、主講教師簡介：

徐琴珍，女，1977 年生，講師，博士，研究方向包括智能信息處理，超聲圖像處理，混合學習方法研究。主持一項國家自然科學基金項目，研究成果以論文形式發表于國際期刊、國内核心期刊以及國際會議上。

五、任課教師信息（包括主講教師）：

Application Form For Opening Graduate Courses

School (Department/Institute)：School of Information Science and Engineering

Course Type: New Open □   Reopen√   Rename □（Please tick in □, the same below）

1. Course Introduction (including teaching goals and requirements) within 300 words:

This course enables the students to comprehend a variety of learning paradigms, algorithms and the research and applications of evolutionary computing in the field of machine learning, and draw on the research results covering concept learning, decision tree learning, artificial neural network knowledge, statistics and estimation theory, Bayesian perspective, computational learning theory, instance-based learning methods, evolutionary computation theory, algorithms for learning sets of rules, analytical learning, combining inductive and analytical learning, and reinforcement learning. The course takes up with clarifying the important links of machine learning involving choosing the training experiences, choosing the target function, choosing a representation for the target function, and choosing a function approximation algorithm. Students who complete this course will have demonstrated the enhanced ability to design a learning system, analyzing problems, and solve problems in the multidisciplinary machine learning field.

2. Teaching Syllabus (including the content of chapters and sections. A sheet can be attached):

(1) The Concept of Machine Learning (Chapter 1)

Including well-posed learning problems, the main links of designing a learning system, and some classic problems and viewpoints in machine learning.

(2)Concept Learning (Chapter 2)

Covering concept learning based on symbolic and logic representation, the general-to-specific ordering over hypotheses, and the need for inductive bias in learning.

(3)Decision Tree and Artificial Neural Network(Chapter 3, 4)

Presenting decision tree learning, the problems of overfitting the training data, the backpropagation algorithm and the general approach of gradient decent for neural network training.

(4)Evaluation Hypotheses(Chapter 5)

Including basic concepts from statistics and estimation theory, evaluating the accuracy of hypotheses using limited samples of data.

(5)Bayesian Learning (Chapter 6)

Including the use of Bayesian analysis to characterize non-Bayesian learning algorithms and specific Bayesian algorithms that explicitly manipulate probabilities.

(6)Computational Learning Theory (Chapter 7)

Covering the Probably Approximately correct learning model, the Mistake-Bound learning model and a discussion of the weighted majority algorithm for combining multiple learning methods.

(7)Instance-based Learning (Chapter 8)

Including k-nearest neighbor learning, locally weighted regression, radial basis function, and case-based reasoning.

(8)Evolutionary Computing (Chapter 9)

Covering the searching strategies of evolutionary computing and implementation methods.

(9)Learning Set of Rules(Chapter 10)

Presenting the theory of sequential covering algorithm and approaches to learning sets of first-order rules.

(10)Analytical Learning, and Combing of Inductive and Analytical Learning

Including purely analytical learning (explanation-based learning), andcombining inductive and analytical learning to improve the accuracy of learned hypotheses.

(11)Reinforcement Learning

Addressing the problems of learning control strategies for autonomous agents, Markov decision process, and Q learning.

3. Teaching Schedule:

Note: 1.Above one, two, and three items are used as teaching Syllabus in Chinese and announced on the Chinese website of Graduate School. The four and five items are preserved in Graduate School.

2. Course terms: Spring, Autumn , and Spring-Autumn term.  

3. The teaching languages for courses: Chinese, English or Chinese-English.

4. Applicable range of discipline: public, first-class discipline, second-class discipline, and third-class discipline.

5. Practice includes: experiment, investigation, research report, etc.

6. Teaching methods: lecture, seminar, practice, etc.

7. Examination for degree courses must be in paper.

8. Teaching material websites are those which have already been announced.

9. Brief introduction of chief lecturer should include: personal information (date of birth, gender, degree achieved, professional title), research direction, teaching and research achievements. (within 100-500 words)

4. Brief Introduction of Chief lecturer:

Qinzhen Xu was born in 1977. She received PhD degree from Signal and Information Processing Major of School of Information Science and Engineering of Southeast University in 2007. She is now an instructor of school of Information Science and Engineering of Southeast University. Her research interests include intelligent information processing, ultrasonic image processing, hybrid learning model. Her research results have been published in international journals, inland journals, and international proceedings.

Luxi Yang was born in 1964. He is now a professor and the director of Digital Signal Processing Division in School of Information Science and Engineering of Southeast University. His major research interests include communication signal processing, MIMO communication system designing, blind signal processing, and space-time signal processing for mobile communications. He has published over 100 journal papers, applied 8 patents for invention and received the first- and second-class prizes of Science and Technology Progress Awards of the State Education Ministry of China for 3 times, and the first-class prizes of Science and Technology Progress Awards of Jiang-su Province of China for 2 times.

5. Lecturer Information (include chief lecturer)