Learn · Build · Impact
Deep Learning has shown outstanding performance in generating consistently accurate forecasts using vast amounts of data, which is hard for humans or traditional modelling.
One area where deep learning can shine is in prediction.
Through this Datathon, you will learn basic to advanced deep learning techniques for forecasting, build your own deep learning algorithm models and make an impact by deploying your models into real world.
No previous programming experience is required.
Your mission is to build an online financial prediction application that predicts the price of a commodity.
We will use a simple trading algorithm to perform trades based on your price predictions. Models with the highest profits over a 8-day period will stand a chance to win this competition.
Your Datathon Journey
LEARN ( 29 June )
8 video sessions (self-paced learning), with online mentorship on building predictive financial applications.
BUILD ( 10 - 19 July )
Participants build their own forecasting model using real-world financial datasets. ai4impact experts provide online mentorship.
DEMO DAY ( 17 July )
Submit a 5-minute video detailing your approach by 17 July 2020.The video will be used to decide if the team qualifies for the final stage of the competition. Judging criteria will be sent to all contestants during this time.
IMPACT ( 20 - 29 July )
If your team is selected to proceed after Demo Day, your model's URL will be used to trigger (virtual) trades of a commodity.
The top 10 teams with the highest profits will be invited to proceed to the Finale. In the event to ties in profits, ai4impact will select the teams to go forward. Details of the trading algorithm will be release closer to the date.
FINALE ( 31 July 2020 )
The top 10 teams will present their solutions to a panel of expert judges who will evaluate their work. This will be done live using zoom. The event will also be broadcast live.
The best teams will stand a chance to become DEEP LEARNING DATATHON 2020 champions!
- The top 3 teams will be presented with Deep Learning Datathon 2020 championship certificates.
- All top 10 teams will be inducted into the Deep Learning Datathon Hall of Fame and receive custom "Deep Learning Datathon 2020 Finalist" T-Shirts.
Frequently Asked Questions
Q: I have no knowledge about programming. Can I still participate? Yes! This competition is accessible to non-programmers. No prior programming experience is needed. Learn and apply the concepts taught and work hard to get a good solution.
Q: Do I have to setup / install anything for the competition? Everything will be accessible through your internet browser (Firefox or Chrome).
Q: Where is the competition held? On the Internet :)
Q: How do I win the competition? We will guide you in building an online financial prediction application. This application will be used to predict the price of a commodity -- exactly which commodit(ies) will be revealed later. We will use a simple trading model (to be revealed later also) to perform trades based on your price predictions.
Q: What is the objective of Deep Learning Datathon 2020? We want you to learn practical deep learning concepts and immediately apply them in building an end-to-end AI prediction application in the area of finance.
Q: Can I just join the training and not the competition? Certainly. But why stop there when you can make a difference?
Q: How large can my team be? You can compete alone or have up to five team members. This datathon is an intense programme with lots to learn. We (highly) recommend you to sign up as a group.
Q: Can I submit multiple models for evaluation? Each team should submit just one model for this competition. However, you are encouraged to tweak your model as the competition progresses.
Q: Do I need to be at a university to join? This competition is open to anyone. Register to learn and compete.
We did not know much about machine learning. We learnt so many things. What's more valuable is the hands-on experience, not just learning the concepts but applying it, it is so important. We highly recommend the Datathon to our peers, just do it! Richard
ai4impact is very helpful in helping me understand deep learning concepts. It provides thorough elaboration on building deep learning models. Solving the problem set given provided us realistic perspectives on applying deep learning techniques in real-world scenarios. Chandana
ai4impact's mission is to help everyone, regardless of age or background, learn AI and build applications. Since its launch in 2019, we have successfully helped hundreds of non-programmers develop useful AI solutions.
We believe AI can help change our lives for the better, and everyone -- yes, that includes you, dear reader! -- should be able to utilize this amazing technology for social good.
To find out more, please visit us at http://ai4impact.org/