Decoded #2

Jul 6, 2018 · Minato, Japan

*Decoded is dedicated to people who are purely interested in AI
*Our seats are limited and we hope that members who truly enjoy the art of AI have access to this event
*This event is free of charge
-----------------
IMPORTANT INFO
1) Come to Roppongi Hills Mori Tower Building Main Entrance (Decoded desk) to get access pass
2) The reception desk opens at 18:30 and closes at 19:30. You CANNOT enter the building after 19:30. Please make sure to get there in time! (The event starts at 19:00)
3) We have a guest list. If we don't have your name, you cannot enter the building!
-----------------

Decoded #2 is coming up! Let's get together and discuss one of the most complex problems around Deep Reinforcement Learning. We have exciting speakers every time!
Enjoy pizza and drinks after the talks :)

Speakers
Fred Almeida (Chief Architect / Founder of Ascent Robotics)
Naoki Shibuya (Research Engineer / Ascent Robotics)
Herve Audren (Robotics Engineer / Ascent Robotics)
Benjamin Smith (ML Engineer / Co-founder of AISCANtech)

-------
Synopsys

*Naoki Shibuya (Research Engineer / Ascent Robotics)*
Reinforcement Learning with World Models

*Herve Audren (Robotics Engineer / Ascent Robotics)*
[Robot Major League]
Although robotic baseball is not yet a thing, in this talk we will try to build a pitching robot. Leveraging both optimal control and reinforcement learning, we will construct, workshop-style, the necessary elements to bring the application to life

*Benjamin Smith (ML Engineer / Co-founder of AI Scan tech)*
AI Scan Tech which is a receipt scanning OCR technology for point of sale receipts, currently targeting US and European receipts. Take a
high resolution picture of a receipt and the API will return all the elements as structured text and numbers. The system was built using Python, Keras and Tensorflow, it utilizes a pipeline of CNNs and AWS GPU P2 instances.
- Overview of AI Scan Tech
- Overview of the receipt OCR model pipeline
- Overview of challenges faced in receipt OCR
Generative Adversarial Networks
- Introduction
- Network Architecture
- Various Types of GAN
- Using GANs in receipt OCR
- Results of Experiments
- Other use cases for GANs
- Recommended libraries, repos and papers to get started

--------

Time table
Fred’s Keynote 19:05-19:15
Naoki 19:15-19:40 + QA19:40 - 19:45
Guest Speaker 19:45-20:05 + QA 20:05-20:10
Break 20:10-20:20
Herve 20:20-20:45 + QA 20:45-20:50
Pizza and Beer time 20:50-21:50
Close 22:00

-----

RSVP opens June 22nd Friday @ 10:00AM
RSVPは6月22日金曜日10:00AMに受付開始です
Everything will be presented in English
英語での進行・発表になります

• What to bring
ID, Business Cards

Event organizers
  • Decoded

    Decoded is an event held by Ascent Robotics, a leading AI startup developing a full stack self-driving vehicle and intelligent robotics.  At our event you can meet the AI researchers and engineers from a range of domains that are changing the face of Japan. Decoded is held bi-monthly and will give you the chance to hear cutting-edge thinking in ML and DRL across a number of domains.  We will even touch on Quantum AI!   The presentations have  wide level of complexity and some math may be needed to understa

    Recent Events
    More

Are you organizing Decoded #2?

Claim the event and start manage its content.

I am the organizer
Social
Rating

based on 0 reviews