Tutorial / Workshop
Normal fee (5,000 JPY) is required if not registered to SCIS&ISIS2014.
Click [HERE] for applicant WITH SCIS&ISIS2014 registration.
Click [HERE] for applicant WITHOUT SCIS&ISIS2014 registration.
*Note : All applicants should go to reception desk at the conference site to get the ticket for tutorial/workshop.
The reception desk will open from noon on Dec. 3rd.
On-site registration is allowed depending on the room capacity, i.e., when the room becomes full, on-site registration will be closed. Thus we recommend you to follow the pre-registration procedure. Normal fee without registration to SCIS&ISIS2014 on-site is 6,000 JPY.
Tutorial on "Almost All about Fuzzy Theory and Systems"
Tutorial 1: December 3, 2014, 13:00-14:00
Fuzzy Sets and Their Operations
Prof. Motohide Umano
Osaka Prefecture University, Japan
Fuzzy sets proposed by Prof. Zadeh in 1965 can represent and manipulate meaning of words in natural languages. We have many applications of fuzzy sets since 1965. In this tutorial, we explain a concept of fuzzy sets and their operations using examples, including 1) what are fuzzy sets and fuzzy relations, 2) three kinds of fuzz-set operations, extension of ordinary set operations, extension of ordinary operations and operations on membership values and 3) type 2 and level 2 fuzzy sets and their operations.
Tutorial 2: December 3, 2014, 14:00-15:00
Fuzzy Control and Fuzzy Decision for Practical Human-Friendly System
Prof. Seiji Yasunobu
University of Tsukuba, Japan
The fuzzy set can handle man's concept with the computer. The fuzzy control (FC) method of using this fuzzy set calculates a nonlinear control command as enhancing the linear PD control. And, predictive fuzzy control (PFC) method decides a gentle operation in consideration of comfort, accuracy, and safety. In this tutorial, I will talk about, 1) FC as easy non-linear control element, and 2) PFC had been applied to practical use with Sendai-Subway Automatic Train Operation system. This system achieved gentle human-friendly train operation.
Tutorial on "Deep Learning"
Tutorial 3: December 3, 2014, 15:15-16:35
Deep Belief Nets and Stacked Auto-Encoders
Professor Kevin Duh
The Nara Institute of Science and Technology (NAIST), Japan
Deep Learning is a family of methods that exploits using deep architectures to learn high-level feature representations from data. Recently, these methods have helped researchers achieve impressive results in various fields within Artificial Intelligence, such as speech recognition, computer vision, and natural language processing. In this tutorial, I will explain the basics of two popular Deep Learning architectures: Deep Belief Networks (DBNs) and Stacked Auto-Encoders (SAE). The goal is to dispel the mystery around Deep Learning and emphasize its simple mathematical foundations. Time permitting, I will also briefly touch upon some applications of these architectures in natural language processing.
Tutorial 4: December 3, 2014, 16:50-18:10
Convolutional Neural Networks and Applications to Computer Vision
Professor Takayuki Okatani
Tohoku University, Japan
Deep learning (i.e., the methods of deep neural networks) has shown its good performances in many areas of AI and is expected to advance further in future. Its applications to computer vision and image recognition have the largest impact on the related research fields. In this tutorial, I will talk about convolutional neural networks, which play the central part particularly in applications to images. In particular, I will explain their origin, mechanism, usage including introduction of publicly available softwares, recent developments, and relation to neuroscience.