Tech SparX - Machine Learning17. května • 18:15 - 20:00 • CA Technologies, V Parku 2308/8, 148 00 Praha 4
Calling all developers, architects, and engineers – are you interested to know more about Machine Learning? Are you looking to hear what we are doing with Machine Learning and to hear about the practical use cases, which can be successful or also failed by practitioners?
Vladimír Kučera – Software Engineer
Vladimír works on storage-related products with the additional strong focus on new technology and specifically machine learning. He has an experience in computer vision and machine learning in his previous roles for more than 4 years. He is as well the author of 4 impacted international journals paper and more than 17 conference papers.
He will tell you about the machine learning historical breakpoints and what’s are the challenges for the future.
Vítězslav Vít VLČEK - Principal Architect
Vítek helps to product management to define and refine features and requirements that a team can work on (http://lifeatca.com/2016/12/07/day-life-ca-principal-software-architect/)
One of many problems in NLP is a sequence to sequence prediction. This problem is effectively solved by encoder-decoder architecture, that is used for example to translate between human languages. He will present the project of his team during CA Hackathon 2017- they tried to predict resource consumption (like CPU usage) based on historical data using above mentioned method.
Jan Samohýl - principal software engineer
Jan has 12 years of experience working on products that support capacity planning and performance analysis of IBM mainframes. His other interests are in functional programming and machine learning, especially the Bayesian methods.
Winner of recent CA Hackathon. His team applied latent Dirichlet allocation (LDA) model to a problem of finding related articles in the CA technical documentation. The LDA is a statistical model that assumes that there are certain topics among given set of articles, and these topics are extracted, in an unsupervised way, based on the co-occurrence of words in those articles. Once the topics are extracted we can compare articles based on the similarity between topics. We used a well-known Python library, GenSim, which implements the inference on LDA (using variational Bayes).
Zdeněk Fiedler – Principal Product Manager
Zdeněk is a machine learning practitioner turning ideas into real deliverables for more than 10 years, currently working in the CA Agile OPS unit focusing on the Application Performance Management.
Machine learning is often a heavy math discussion about algorithms or structuring neural networks. However, the true key to any successful machine learning application is data.
He will briefly talk about an impact the training data have on the usability of one of machine learning applications CA is working on.
Beer, drink & refreshment will be provided.