From 21st to 23rd April I have taken part in the Machine Learning Conference in Prague, an event dedicated to applications of the technologies related with machine learning. Overall, the conference provided a full 14 hours of hands-on information with 600 attendees, amongst which three people from the team of Konica Minolta Laboratory Europe also had the chance to listen a wide variety of inspiring talks.
Among the topics discussed in the three days of the conference two were mentioned frequently in several of the presentations: the gradual learning and the deep learning. The former is an evolution of reinforcement learning that is a development grounded on the most renowned concept of deep learning. Deep learning has demonstrated to be widely used, not only as the basis for self-improvement of several gradual learning techniques (as described by GoodAI on a workshop and in the presentation by Kristinn R. Thórisson), but it can be applied also in several different application, such as for instance:
- Malware detection (Avast)
- Image search (Seznam.cz)
- Analysis of sounds for malfunctions of machines (Neuron soundware).
Many of the topics covered by the conference were not simply restricted to technical presentations: we had the chance to hear about the development of Machine Learning start-ups from the perspective of Data Collective VC, questions of morality and the overall risks of the development of artificial intelligence.
I had the pleasure to describe how to quickly prototype machine learning systems on the basis of the lessons we have learnt in Konica Minolta within a project related to the predictive maintenance of Multifunctional Printers. My presentation focused on providing general principles about the process of prototyping as well as covering some of the technical details about the project itself. I am glad that this mixed presentation was met with a positive feedback from the audience, and I would like to thank all the people who asked me so many interesting questions.
At the end of the conference selected speakers from Facebook, The New York Times, Data Collective VC, and the Icelandic Institute for Intelligent Machines discussed the emerging problems of Machine Learning, such as the problems related with fake news, or how the idea of Open Data can impact on machine learning development. An interesting summary of the event is available in the blog from Simon Dittlman.
I would like to thank the organizers for a very inspiring event, where I met numbers of great people and I had the chance to present the story of Predictive Maintenance at Konica Minolta. The MLPrague 2018 has been already announced, so I hope to see you next year in Prague.