AAI-IST2017A | Deep Learning JP

Sorry, this entry is only available in Japanese. For the sake of viewer convenience, the content is shown below in the alternative language. You may click the link to switch the active language.

Frontier Artificial Intelligence II 2017（The course of Graduate School of The University of Tokyo）

The emergence of Deep Learning technology has dramatically improved AI technology, and had a big impact on diverse industry and social institution. In the near future, every industry and its types should be influenced by AI technology. Frontier artificial intelligence II provides a project-style lecture intended acquisition of more practical ability of research and development for the learners who have a basic understanding of Deep Learning and can construct the models. Based on the idea “Practice makes perfect”, we aim to let you acquire the technique through the exercise. In exercise, you can concentrate on the point of the lecture by using our specific development platform “ilect.net” : enables you to code in Python with GPU on a browser（The detail of iLect is here）.

Every Tuesday fourth class（from 2:55 PM to 4:40 PM）
Place：Faculty of Engineering Building 2, Second floor lecture room 221, The University of Tokyo
Target：Graduate students of the University of Tokyo（Notes：The number of participants will be limited to about 50 people due to the limitation of the exercise system.

application
Q & A

Program syllabus

（The contents are being adjusted, may be changed.)

	date	title	contents
1.	2017/09/26	Introduction/Guidance	– Summarization of the spring lecture – Introduction of the autumn lecture, this year’s highlights, assessment of achievement, etc…
2.	2017/10/03	Advanced image recognition１	Object recognition, CNN, Transfer Learning, Fine-Tuning, Introduction of open source softwares, Reuse of pre-trained network,
3.	2017/10/10	Reinforcement１	Outline of reinforcement learning – problem establishment in reinforcement learning – value function based approach and policy based approach – Introduction of Chainer, ChainerRL
4.	2017/10/17	Generative models１	Outline of generative models, RBM
5.	2017/10/24	Big data and Deep Learning	HPC, GPU, Profilers, Database and Deep Learning
6.	2017/10/31	Team development（methodology, teaming）	Team development, Git and workflow, the project, precautions, etc…
7.	2017/11/07	Advanced image recognition２	Region proposal, Fast/Faster RCNN, RPN, Single-shot architecture
	※2017/11/14	（no lectures）
8.	2017/11/21	Reinforcement learning２	Value function based approach – DQN and respective researches – Expansion of continuous actions
9.	2017/11/28	Generative models２	Variational Auto Encoder (VAE) and respective topics
10.	2017/12/05	Advanced image recognition３	Segmentation
11.	2017/12/12	Generative models３	Generative Adversarial Networks (GAN) and respective topics
12.	2017/12/19	Reinforcement learning３	Policy based approach – Policy gradient – A3C, TRPO, PPO – Case studies (Acrobot, Atari, etc…)
13.	2018/1/9	Interim report	Lightning talk style
	2018/1/16	a supplementary class
	2018/1/20	Copyright © 2017 Deep Learning JP. All Rights Reserved.