How to Transfer ML model to Java and Got an Inference Speed Boost

Vitalii Tsymbaliuk visited NIX MultiConf #5 as an experienced Data Scientist to talk about the practical case – How to Transfer ML model to Java and Got an Inference Speed Boost

Vitalii shares first-hand experience of going through a range of underlying processes, including binary classifications, Scikit-learn nuances, as well as specifics of a number of other tools to achieve the inference speed boost result. The presentation includes detailed highlights of all the practical tasks, demonstrations of results, and major takeaways.

NIX MultiConf #5 takes another round to bring together specialists, enthusiasts, and entrepreneurs from across IT niches and discuss all that’s on today’s global IT market agenda. Organized by one of Ukraine’s leading software agencies, the event gives you up-to-date knowledge and boosts expertise.

Get your in-depth insights from our regular NIX MultiConf #5 video reports and visit our website for more info.

You may also get a free consultation with our experts.

----------

Our website

Follow us here

#NIX_United #NIXMultiConf #Data

----------

00:00 Intro
0:01:41 Presentation of the topic
0:00:58 Serving LGBM in Java
0:01:14 About speaker
0:02:12 What this lecture reveals
0:03:45 Binary classification
0:05:28 Scikit-learn
0:06:19 XGBoost
0:07:20 LightGBM
0:07:59 Serving
0:10:55 What is PMML?
0:11:57 Set of libs to compare
0:14:19 Demo time
0:15:22 Practical part
0:15:54 Python code
0:22:02 Java application
0:32:40 Demo summary
0:34:22 Performance comparison
0:35:45 Lessons learned
0:40:12 Q&A session