The qualification round of DeepHack.Turing, an international hackathon and scientific school with a focus on artificial intelligence and deep learning, has begun. It will be followed by an on-site competition to be held 24-30 July at MIPT. Those taking part in the hackathon will gain valuable experience, and the winning team will travel to Los Angeles to participate in the Conference on Neural Information Processing Systems (NIPS).
The hackathon organizers will have participants tackle the problem of automating the Turing test, a classic way of telling if a machine can think. Conversation agents have become an integral part of our daily lives. They currently have a wide range of practical applications, including data collection and customer service. But as of now, there are no effective methods available for automatic evaluation of dialogue systems. That is why the event organizers are bringing together talented students and young researchers, as well as information technology professionals and natural language processing specialists to develop bots capable of differentiating between human- and model-generated responses in a dialogue.
“Our goal is to take communication between a person and a machine to a new level,” says Mikhail Burtsev, head of the Laboratory of Neural Networks and Deep Learning at MIPT. “In the early 2017, we launched the three-year iPavlov project, which aims to improve the ability of machines to engage in meaningful dialogue with humans. To do this, Russian scientists are joining forces with their colleagues from abroad. Our partners include the Université de Montréal, McGill University, Carnegie Mellon University, and many others. At the end of the year, the final round of the competition between bots — the Conversational Intelligence Challenge — will be held at the NIPS conference in California. The DeepHack.Turing school and hackathon will contribute to the qualification round of that contest: The participants will test their own bots and then improve their models based on test results.”
Olga Kairova, deputy head of the laboratory, explains that the hackathon is part of the iPavlov project and technologies developed within this project will be made publicly available. She adds: “Our contests are a way of always being at the forefront of dialogue systems development. This is not just about research: The goal of project iPavlov is to find business applications for the technologies we create. That is why we are making them available as an open-access library of algorithms and models, all ready for use in applied dialogue solutions. To find out what kinds of technologies are in demand, we are contacting companies interested in dialogue bot implementation and inviting them to join us and set the requirements for the library, right at the outset.”
For the past several years, MIPT has been a hub for talented young researchers working on some of the most exciting problems in the field of AI. In their effort, they are aided by leading scientists from top research centers such as DeepMind, OpenAI, Facebook AI Research, Google, and Yandex.
DeepHack.Turing is the fourth in a series of DeepHack events focusing on neural networks. This time the hackathon is part of the iPavlov project. The event received organizational support from the working group of the NeuroNet National Technology Initiative and OpenDataScience. The prime sponsors of DeepHack.Turing are MIPT and Sberbank. Among the partners of DeepHack.Turing and the Conversational Intelligence Challenge is Flint Capital, an international venture capital fund with considerable AI and machine learning expertise.
“Flint Capital decided to support this initiative to foster international cooperation in the field,” says Andrew Gershfeld, a partner at Flint Capital. “Our purpose is to help Russian developers enter the global market with their products and technologies.”
Detailed information is available at: http://deephack.me
To find out more about participating in the contest, contact organizers at: email@example.com