Radicalbit at Big Data Conference Europe 2023

Nov 29, 2023 | by Radicalbit, Technology

The days between November 21 and 24 were truly exciting as we had the honour of presenting a live talk at the Big Data Conference Europe 2023 in Vilnius, Lithuania.

This experience wasn’t just a deep dive into cutting-edge topics but also an immersion into Lithuanian culture. Despite the cold weather, a warm welcome awaited all the speakers in the form of a delicious dinner featuring typical dishes, providing a great way to refuel our minds before the upcoming days.

The following is the event best-of reported by our Senior Data Scientist Mauro Mariniello.

The venue

With full bellies and warm clothes, the conference finally began on November 22.

The first day featured numerous panel discussions and talks, covering intriguing topics such as the impact of LLMs and Generative AI on the world.  The presentations offered various solutions for integrating these technologies across a wide range of use cases. A couple of talks on Vector Databases, gaining popularity thanks to LLMs, were particularly noteworthy.

These databases are used to store and query the embedded vectors created by these models, enabling the creation of AI agents capable of recovering and remembering vast amounts of information.

The day of the talk

Our day came on November 23 at 2:15 PM when the stage in Hall 5 eagerly awaited my presentation. I spoke on a topic dear to Radicalbit – Online Machine Learning. The speech began with an explanation of this learning paradigm, emphasising the main differences between Online Learning (training the model using a data stream) and Offline Learning (training the model using a training set stored in batches).

A particular focus was placed on how these methods can be affected when a Concept Drift (changes in the relationship between features and target) occurs, highlighting the consequences in scenarios like Offline Learning without Concept Drift, Online Learning without Concept Drift, Offline Learning with Concept Drift, and Online Learning with Concept Drift.

I then had the honour of presenting the brilliant work accomplished by Lorenzo D’Agostino during his internship at Radicalbit. He worked on a thesis titled “Analysis and Benchmarking of algorithms for Streaming Machine Learning and technologies for edge deployment,” in which Lorenzo implemented an Online XGBoost (trained using a data stream). The presentation covered the achieved outcomes and Lorenzo’s analysis of cloud and edge deployment.

big data conference vilnius 2023

During my talk, I provided the audience with a summary of this work, and starting from this model, I demonstrated how to leverage Radicalbit’s MLOps platform to deploy it and monitor its performance over time. The day wrapped up with a lively party at “ACTION! by APOLLO,” featuring beer, bowling, and board games that engaged both speakers and attendees.

November 24 marked the last day of the conference, during which I had the pleasure of listening to two talks and participating in a panel discussion. The first talk focused on a fascinating LLM app. The second discussed threats and opportunities in using Kafka, and the panel discussion delved into the role of Agile methodologies in data analytics.

A necessary consideration concerns the most discussed theme during the conference, which, as expected, was represented by generative AI and particularly by LLMs. It is surprising to see how many different people are working to build applications that leverage these powerful models in various ingenious ways. Noteworthy, however, is the role that “data” has played. Many speakers have dedicated time and words to the concepts of data security, privacy, and maintenance, which I think are as important as data is the lifeblood of any AI model.

In conclusion, the Big Data Conference was an exceptional event, genuinely focused on speakers and not solely on companies. We had the opportunity to meet numerous professionals, laying the foundations for future collaborations and discovering a vibrant atmosphere between the Baltic and East European regions.

The last thing to say, ačiū, Lietuva!

big data conference europe
big data conference europe 2023 talk
How RAGs Help Mitigate LLM Hallucinations: 5 Use Cases

How RAGs Help Mitigate LLM Hallucinations: 5 Use Cases

RAG constitutes one of the most exciting developments of artificial intelligence. The technique allows combining Large Language Models (LLMs) with external knowledge bases in order to increase the accuracy and reliability of generated answers. In other words, it means grounding the generative AI model with facts and information that were not previously employed to train the model.

How MLOps accelerates AI Model Deployment

How MLOps accelerates AI Model Deployment

MLOps is the bridge between machine learning and operations. A combination of methodology, tools and processes, it streamlines and automatizes the ML model lifecycle management, integrating ML workflows, pipeline and automation, continuous delivery and observability. This article explains how MLOps can be the conductor that makes all the difference.