Численные методы обучения по прецедентам (практика, В.В. Стрижов)/Basic schedule

Материал из MachineLearning.

Перейти к: навигация, поиск


The course outline: lecture plan and the students' tasks

Lecture coverage Task to complete Result
1 Introduction into the course, the course format: the course syllabus, grading system, consultations setup. The lecture is organized in a form of conversation, encouraging the students to think on their own motivation to complete the course. To download all necessary instruments and create accounts for the resources, utilized during the course; to read the introductory materials.
2 Demonstration of the interfaces. Introduction to the basic instruments involved in the course routine. Presentation of test programming problems. To chose and solve solve a test problem from the list. To upload the matlab code and the illustrations, written for the problem, into an appropriate directory in the repository. Test: the research title and a link to the research folder on SourceForge are published on the groups page on machinelearning.ru
3 Presentation of the research problems: the lecturer provides concise commentary on each problem. The basic principles of finding literature, writing annotation and introduction to a research paper. To find a number of papers on the selected topic for the list of references, to write an annotation and introduction with a literature overview based on the collected references. Annotation: a passage of approximately 600 characters, Introduction: the motivation to the research, literature overview and suggested approaches (about 1-2 pages), Litreture: a list references issued as a bib-file.
4 The analysis of the previous task: the lecturer looks through randomly chosen paper drafts, emphasizes most common mistakes made by students and gives advices on how to avoid such mistakes. The basics of mathematical problem statement. To write the problem statement and perform a basic computational experiment, supported with the initial result analysis. Statement: problem statement (0.5-1 pages), Basic code, Report: the code and a concise report on the performance of the basic algorithm are uploaded to the repository.
5 The analysis of the previous task. Recommendations on illustrating the experiment. To conduct a full version of the computation experiment including the newly suggested methods. Designing the experiment, students should consider the results of the previous experiment. To prepare a report. Code, Visualisation: the supplemented code and the visualizing pictures are uploaded.
6 The students are reporting (Middle) the results of working on the paper: problem statement, literature overview the results of basic and full computational experiments. To provide the description of the suggested algorithm and the computational experiment. The results of the experiments are demonstrated with the tables and figures illustrating the error analysis and the quality of the algorithms performance. Theory: the section with theoretical grounding of the suggested approach is added to the paper draft.
7 The structure analysis: distinguishing between the problem statement, the description of the algorithms and computational experiment. To revise the theoretical part and the description of the computational experiment. To provide a clear explanation of each figure, and write a conclusion to the paper. Document: a complete draft of the paper.
8 More ways of presenting the results of the computational experiment. Error analysis. To finish the computational experiment; to enhance the description of the computational experiment and error analysis. Error: explicit error analysis added to the description of the computational experiments.
9 Final report (Final): the students present the results of the semester work. To fix the paper according to the revisions made during the report. cHeck: the paper is completed.
Личные инструменты