The Data Science UA Conference 2018 was organized in Kiev to bring together experts in Machine Learning, Data Analytics, BI, Data Science, AI for a day-long discussion on how data is changing the world today and how the digital landscape will look like tomorrow. Technology specialists from different countries came to discuss new methods, tools, and real cases of applying technologies for improving business efficiency. Our Db Devs team was among the participants.
During the conference day many topics related to data science were covered and plenty of issues were raised and discussed. Data specialists have been explaining a wide range of complex concepts, including: algorithms, used for playing games (Chess, Go); architecture of large applications for working with Big Data; the problem of complex data structures and their applications within transporting systems; mathematical algorithms for resolving problems of car positioning using GPS and other sensors; algorithms used for data visualization and many others.
The range of topics Db Devs team found particularly interesting and the brief overview of the concepts discussed is outlined below.
What Do AlphaGo and AlphaZero Do, Exactly? Deep Reinforcement Learning.
In this presentation the speaker covered the approaches of using neural networks in application to logical games such as chess and ancient Chinese game Go.
Building Data Science Platform as a Main Component of Enterprise Digital Transformation.
The problem of storing and processing massive data amounts at multiple servers, intercommunication between parts of the system, and processing of multiple requests was highlighted in this speech. Also, advantages and disadvantages of different technologies were explained.
The guys agreed that this part of the conference seemed to them the most informative and the main ideas given during the speech they found fully applicable for building huge scalable data storages.
Latent Representations and Variational Autoencoders.
During this part of the presentation the speaker shared the information on how to use neural networks for compressing and decompressing redundant digital data. Predicting the Unpredictable. Agent-based Modeling.
In this speech the presenter highlighted the problem of complex systems and explained it by resolving a mathematical model, based on several simple rules. The difference between complex and complicated systems was also explained.
Fusion of Visual Recognition Results and Sensors Data using Unscented Kalman Filtering.
The presenter explained how the problem of a car positioning on the map using the geographic coordinate system can be resolved with the help of Kalman Filtering. The practical application of GPS navigation and car computers was also addressed.
The final part of the conference was devoted to the discussion of a current state of Data Science in Ukraine and in the world as well as the future of a Big Data and Machine Learning. Db Devs team say, that many topics addressed were indeed of high interest to them and touched many practical aspects. However, unfortunately, despite the name “The Data Science Conference”, not all the topics were fully relevant to the overall Data Science concept. Most of the topics were related to narrow problems that speakers faced while working on their own projects.
The overall impression the Data Science UA Conference 2018 left was that Data Science in Ukraine is a very young direction which just started gaining momentum. Due to a shortage of experience in classical development approaches, most of the Big Data projects are being developed intuitively and Big Data is still most commonly being perceived as just a huge amount of some digital information.
However, Db Devs team says, that looking at the success of Machine Learning and Data Science in different spheres of our life like marketing, healthcare, games, science, many Ukrainian specialists are gradually turning into these directions as many of them understand that these technologies will soon change our future.