講演会開催(2017年11月21日)のご案内

IEEE Computer Society Fukuoka Chapterでは、下記の講演会を開催いたします。
奮ってご参加ください。

【主題】Questioning Common Deep Learning Practices and Advanced Applications

【プログラム】
1. Marcus Liwicki: Dynamic Cortex Memory and Applications
2. Fotini Simistira: OCR with LSTM using ocropus & ocropy
3. Michele Alberti: Initialisation of networks using PCA/LDA
4. Vinaychandran Pondenkandath: Transfer Learning

All speakers are from University of Fribourg, Switzerland

【日時】平成 29年11月21日(火)10:30~12:00

【場所】九州大学伊都キャンパスウエスト2号館3F
システム情報科学府第3講義室(W2-312)

【お問い合わせ】
九州大学大学院システム情報科学研究院
内田誠一 uchida _at_ ait.kyushu-u.ac.jp (_at_を@に置き換えてください)

【講演概要】

In this presentation we will dig a bit deeper into deep learning and
present some unconventional methods and successful applications.
First, an advanced LSTM(we assume the audience knowing LSTM) model
called Dynamic Cortex Memory will be presented and analyzed.
Next, we question typical initialization (random weights) and
optimization strategies (Adam) and show alternatives. Also we give
an explanation why ReLU should not be used. Finally, we show new ideas
of Transfer learning and successful applications in Document Processing,
Robotics, and Sensor Data Analysis.

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

*