Fundamentals of Artificial Intelligence-E2
| Numbering Code | U-LAS30 20036 LE10 | Year/Term | 2022 ・ Second semester | |
|---|---|---|---|---|
| Number of Credits | 2 | Course Type | Lecture | |
| Target Year | All students | Target Student | For all majors | |
| Language | English | Day/Period | Mon.2 | |
| Instructor name | CHU, Chenhui (Graduate School of Informatics Program-Specific Associate Professor) | |||
| Outline and Purpose of the Course | Recent development in artificial intelligence techniques (AI), in particular the set of techniques commonly referred to as “deep learning,” has significantly increased the number of tasks that computers can solve easily. This leads to a current explosion in the use of AI: chatbots helping users on commercial websites, self-driving cars, automatic translation, automatic photos tagging, etc. It is of course not possible to introduce all aspects of AI in one semester, but this course will attempt to give a sufficiently detailed explanation of at least a few of the most AI common techniques. We will focus on supervised machine learning in general and deep learning in particular. One goal will be to give practical and working knowledge to students, so that they can apply what they learned to at least some simple tasks. | |||
| Course Goals | Students will have a good understanding of simple supervised machine learning techniques, and be able to implement and use some for automatic classification tasks. | |||
| Schedule and Contents |
1. Overview of Artificial Intelligence and this Course (1 week) This will give a “big picture” description of the field of AI. We would first discuss some common applications of AI: game AI, chatbots, machine translation, and automation (self-driving vehicles, robots) etc. Then we would discuss paradigm of machine learning (supervised, semi-supervised, and unsupervised) and give an overview of this course. 2. Review of Mathematics Concepts (3 weeks) Firstly, we will spend one lecture studying the basics of the Python programming language. Then, we will review some of the mathematics concepts that are the most necessary for the understanding of AI methods. In particular, we will review essential notions of calculus and optimization (derivative, numerical methods for finding a minimum), vector and matrix. Finally, we will learn how to minimize a function with stochastic gradient descent and implement it in Python. 3. Basic Supervised Machine Learning (3 weeks) Focusing on simple tasks of simple/multiple linear regression and classification, we introduce the terminology and basics of machine learning: defining a parameterized model, defining a loss, train the model parameters by minimizing the loss. We will also introduce how to implement simple/multiple linear regression in Python. 4. Deep Learning (3 weeks) We will first introduce the basic ideas of deep learning neural networks. Then we will study the architecture of neural networks and the back-propagation algorithm for optimizing neural networks. Finally, we will look at one of the most important type of neural network architectures: feed-forward with fully-connected layers and study how to implement them using the deep learning framework Chainer. 5. Computer Vision and Natural Language Processing (4 weeks) We will first give a brief introduction of computer vision: what is an image for a computer; what are convolution layers. Then we will study how to build an object recognition neural network with convolution layers, max-pooling layers and fully-connected layers. Next, we will implement and train a real object recognition neural network in Chainer. Finally, we will have a quick look at recurrent architectures and how they are used to process text. As a final application, students will be asked to solve a real problem in their studies using the models (either basic supervised machine learning or deep learning) introduced in this course. 10. Feedback (1 week) |
|||
| Evaluation Methods and Policy | Evaluation is based on class participation (15%), mini reports and exercises (60%), and the final report of solving a real problem in students’ studies using the models learned in this course (25%). | |||
| Course Requirements | None | |||
| Study outside of Class (preparation and review) | The instructor expects students to spend about 60 minutes after each class to review the content. Some practical exercises will also be given at the end of some lectures so as to let the students see how much of the content they do understand practically. | |||
| Textbooks | Textbooks/References | Lecture handouts will be provided in the class. | ||
| References, etc. | Deep Learning, Ian Goodfellow, Yoshua Bengio and Aaron Courville, (The MIT Press), ISBN:978-0262035613, 2016 | |||
