BLOG CUBE SYSTEM Official Blog

  1. HOME
  2. BLOG
  3. Our team at CDLE HACKATHON 2020 ...

AI

Our team plays an active part in CDLE HACKATHON 2020!

2021.01.19
Text:

アイデア部門と予測性能部門の2部門で受賞!


The work of a team of our employees entered the "CDLE HACKATHON2020" sponsored by the Japan Deep Learning Association (JDLA) and received the following awards!
・ Idea category: Received a corporate award
・ Predictive performance category: Winning
The award-winning members are participating members of the "Top Runner Program," an in-house activity that studies AI, blockchain, CX / UX, and so on.

About CDLE HACKATHON 2020

"CDLE HACKATHON 2020" is an annual hackathon (a competition on AI technology) run by JDLA. Participation qualifications are those who have passed the G test and E qualification, which are qualification tests (CDLE members). The 2020 issue was implemented with the following two issue
 Idea section: NEO e-learning utilizing DL (Deep Learning)
 Prediction performance department: Meteorological prediction based on image data

What is the Top Runner Program?

At our company, there are currently three types of top runners, the AI Top Runner Program, the AWS (Amazon Web Services) Top Runner Program, and the CX / UX (Customer Experience / user ・ We carry out programs and provide various environments for participating employees, such as participation in training, events / seminars.

Award-winning work

The award-winning works will be the following works.
<Idea section>
The work submitted at the ideathon of "Neo e-learning" using DL won the award!
 Award-winning work: "Technical writing training by deep learning"
 Comments from the selection committee
"In the e-learning so far, technical writing is a field that is necessary as society
On the other hand, there is a niche in how to write documents using DL and how to create documents unique to the industry, and I would like to introduce it immediately.
That was the decision to win the award. "
 Interview with Winners (external link):https://digital.baycurrent.co.jp/feature/archives/34

<Predictive performance section>
The task was to build an AI model that predicts the movement of clouds 24 hours later based on the satellite images (cloud movements) of the past 96 hours based on the satellite images provided by Weather News.
SSIM (Structual Similarity) was adopted for the evaluation, and the similarity between the correct image and the predicted image was calculated from the viewpoints of brightness average, contrast, structure, etc., and the ranking was calculated.
As a result, we won a prize!

We will continue to contribute to the development of customer business, such as the creation of Services, by actively working on cutting-edge technologies.

Related keywords

Latest latest article

Category Article category

Tag Popular tags