Buy tickets

Sava Centar, Belgrade

20-24th November 2023

DSC SCHEDULE

Download Preliminary Schedule
DATA & AI TALKS
BREAKOUT SESSIONS
CO-LOCATED EVENTS

Tech-Tutorials

Stream 1

LLM applications with LangChain

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
Library Demonstration
Intermediate
“The goal of this tutorial is to familiarize the participants with LangChain library, a Python framework for building applications based on large language models such as ChatGPT. We will introduce main LangChain concepts such as agents, chains and memory. To demonstrate the capabilities of LangChain, we will solve two problems that we might encounter in practice: a QA chatbot for answering questions over a collection of internal documents and an assistant for querying relational databases using natural language.”
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
09:00

 —

11:00

LLM applications with LangChain

Andrej Miscic & Luka Vranjes
Chief Scientist @ Valira AI, Data Scientist @ Valira AI

Reimagining DataFrames: An Introduction to Graphs and GNNs with PyTorch Geometric

Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Data Wrangling, Data Mining, Edge Computing
Career Path, Library Demonstration
Beginner to Intermediate
Discovering meaningful connections within data can significantly impact problem-solving. Leveraging this notion, DataFrames or tabular data can be transformed into graphs, where interconnected data points enhance predictive performance. Graph Neural Networks (GNNs) have emerged as powerful tools for machine learning on graphs. This tutorial delves into the fundamentals of graph structures and GNNs, combining theoretical concepts with practical implementation using PyTorch Geometric and NetworkX. Participants will learn to convert a tabular dataset into a graph and train a GNN model for a specific task, gaining valuable insights into the potential of these techniques.
Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Data Wrangling, Data Mining, Edge Computing
11:15

 —

13:15

Reimagining DataFrames: An Introduction to Graphs and GNNs with PyTorch Geometric

Darja Cvetkovic
Research Assistant @ Institute of Physics Belgrade

Build your own Generative AI Chatbot with watsonx Assistant

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
Tool Showcase
Intermediate
In the tutorial, I will showcase how anyone can create a chatbot that harnesses the power of Generative AI. In the tutorial, we will build the dialog from scratch, I’ll showcase how to perform webhooks and how to connect it to an open-source Large Language Model.
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
13:30

 —

15:30

Build your own Generative AI Chatbot with watsonx Assistant

Erik Ternav
Technology Engineer @ IBM Slovenia

Optimizing workloads with Transformers and Diffusers

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation, Image Classification, Image Processing, Machine Vision, Computer Vision
Library Demonstration, Tool Showcase
Beginner to Intermediate
In today’s data-driven world, the utilization of Transformers and Diffusers, two powerful deep learning architectures, holds immense potential for optimizing a wide range of workloads. This presentation explores the synergy between these models, their applications in natural language processing, computer vision, and beyond. Discover how the unique properties of Transformers and Diffusers can revolutionize workload optimization, improving efficiency and performance in various domains. My colleague Daniel Socek will help me with holding the tutorial.
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation, Image Classification, Image Processing, Machine Vision, Computer Vision
15:45

 —

17:15

Optimizing workloads with Transformers and Diffusers

Stefan Dejanovic
CTO @ GapApp

Stream 2

Deep Learning Techniques for Monocular Depth Estimation

Image Classification, Image Processing, Machine Vision, Computer Vision, Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Transfer Learning, Adaptive Binning, Vision Transformers
Career Path, Tool Showcase
Intermediate
Monocular depth estimation is mathematically an ill-posed problem as infinitely many 3D scenes can be projected to the same 2D plane. This tutorial offers exploration into the dynamic realm of monocular depth estimation, spotlighting influential methodologies from recent years. We’ll journey from the transformative role of transfer learning, delve into the practicality of adaptive binning techniques, and investigate the innovative application of Vision Transformers. Pivotal to our discourse, we’ll also unveil the groundbreaking DINOv2 model by Meta AI, which harnesses the recent breakthroughs from NLP.
Image Classification, Image Processing, Machine Vision, Computer Vision, Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Transfer Learning, Adaptive Binning, Vision Transformers
09:00

 —

11:00

Deep Learning Techniques for Monocular Depth Estimation

Vladimir Kovacevic
Teaching Associate @ Faculty of Sciences, University of Novi Sad

Productionizing Jupyter Notebooks

Operationalizing Data Science, Model Deployment, Notebook Production, Data Science Tooling
Tool Showcase
Intermediate
Jupyter notebooks play a crucial role in the development stage for data analytics, machine learning, and other data-driven applications. However, moving them into a production environment often leads to a complex set of challenges. Challenges related to production readiness, version control, testing, reproducibility, and modularity are common. In this talk, we will delve into the specifics of these challenges and demonstrate how our tool, Versatile Data Kit (VDK), effectively addresses them. From deploying production-relevant code to ensuring linear execution for reproducibility, we will provide insights into practical solutions that could enhance efficiency in your data analytics and machine learning workflows. This session is aimed at those looking to understand the complexities of productionizing notebooks and explore potential methods to overcome these challenges.
Operationalizing Data Science, Model Deployment, Notebook Production, Data Science Tooling
11:15

 —

13:15

Productionizing Jupyter Notebooks

Antoni Ivanov & Duygu Hasan
Staff Engineer @VMware, Member of Technical Staff @ VMware

Reinforcement learning though gamification (Learning how to learn with AI and videogames)

Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Intelligent agent developemnt
Career Path
Beginner to Intermediate
In this tutorial, participants will learn the basics of reinforcement learning by creating intelligent agents powered by artificial neural networks in a fun and interactive way. Participants will be guided towards creating a simple game word occupied by a player agent. Once an environment is set up and the concepts behind a game world are defined and explained participants will be instructed on how to create a population of intelligent agents that, through an iterative process, learn how to beat the created game efficiently. Complex concepts such as objective evaluation functions, neural networks will be covered interactively and visually.
Machine Learning, Deep Learning, GNNs, CNNs, RNNs, Intelligent agent developemnt
13:30

 —

15:30

Reinforcement learning though gamification (Learning how to learn with AI and videogames)

Luka Jovanovic
Master Student @ Singidunum University

The Synthetic Data Advantage: How It Helps You to Get Both AI Privacy & Explainability

Machine Learning, Deep Learning, GNNs, CNNs, RNNs
Career Path
Beginner to Intermediate
Synthetic Data is seen as the key enabler for AI privacy that helps thousands of data scientists to stop waiting ages for data access – and finally, get to work. Curious how? Join this interactive session, a DSC Europe 2022 favorite!

You’ll discover:
-What AI-generated synthetic data is
-How major insurance, telecom, and banking giants use it
-How it safeguards privacy and unlocks data access

More excitingly, you’ll gain hands-on synthetic data skills:
-Create your own, high-quality synthetic data
-Assessing synthetic data‘s utility for machine learning
-Leveraging synthetic data for Explainable AI (without sacrificing privacy!)
-Plus, a bonus: Smart Imputation to save time and improve quality during data pre-processing
Machine Learning, Deep Learning, GNNs, CNNs, RNNs
15:45

 —

17:15

The Synthetic Data Advantage: How It Helps You to Get Both AI Privacy & Explainability

Alexandra Ebert
Chief Trust Officer & LinkedIn Learning Instructor @ MOSTLY AI

Stream  3

Lasso and Ridge regression models in R: Introduction

Machine Learning, Deep Learning, GNNs, CNNs, RNNs
Career Path, Tool Showcase
Intermediate
Linear regression models are practical because they are understandable, relatively easy to implement and very simple, considering that they are based on fitting a regression line through the data. If the model is statistically significant it can be used for prediction and estimation, but does it always work? No, sometimes the model is overfitted to the sampled data, and cannot handle new instances. What comes as a solution is regularisation which reduces over-fitting, increases model reliability and prediction and estimation accuracy. Two types of regression with regularisation will be introduced, both theoretically and practically: Lasso and Ridge regression.
Machine Learning, Deep Learning, GNNs, CNNs, RNNs
09:00

 —

11:00

Lasso and Ridge regression models in R: Introduction

Milica Maricic
Assistant Professor @ University of Belgrade, Faculty of Organizational Sciences

Fall in love with the data transformation – DBT as your new weapon of choice

Data Wrangling, Data Mining, Edge Computing
Tool Showcase
Beginner to Intermediate
“In this comprehensive tutorial, a GitLab Data veteran will show the power of Data Base Transformer, also known as DBT, a rising star in the Data World. You will get basic knowledge about a use case where, how and why to use DBT. This is an ideal chance to catch the best mutual venture for SQL and Python. Why this is important: Whether you are a Data veteran or you just started your professional journey, this session will bring a smile to your face and expose you to how to leverage data transformation and make your life easier. DBT is a “”swiss-knife”” solution among Data Transformation tools. If you like Open Source tools, Python, SQL and Data, this is the tutorial you shouldn’t miss. “
Data Wrangling, Data Mining, Edge Computing
11:15

 —

13:15

Fall in love with the data transformation – DBT as your new weapon of choice

Radovan Bacovic
Staff Data Engineer @ GitLab

Storytelling through data

Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
Tool Showcase
Beginner
“How to generate the best insights from the data?

Storytelling through data is an art that transforms raw information into compelling narratives, offering invaluable insights. To generate the best insights, it’s crucial to adopt a structured approach. Begin by understanding the data’s context, establishing clear objectives, and identifying key variables. Utilize visualization tools to create meaningful representations, enhancing comprehension.

Effective storytelling involves weaving a coherent narrative that guides the audience through the data’s journey. Integrate human elements into the analysis, making the information relatable and engaging. Craft a storyline that not only highlights trends and patterns but also addresses the “”why”” behind the data. Contextualizing data helps uncover actionable insights, empowering decision-makers.

Furthermore, simplicity is key. Present complex findings in a digestible manner, catering to diverse audiences. Ultimately, mastering the art of storytelling through data is a dynamic process, merging analytical prowess with communicative finesse to extract the richest insights from the information at hand. “
Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
13:30

 —

15:30

Storytelling through data

Milica Scepanovic
Head of consulting @ Menadzment Centar Beograd

Identification of Building Thermal Model – Using Data

Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
Career Path
Beginner to Intermediate
First step to applying control algorithm to anything is making a model of a process you’re trying to control. In buildings, first you have to obtain measurement data for at least one year (or use a simulation software that will generate the data using the pecise info on materials of the building). Then, you have to run the data through complex math model that will give you model parameters called Unscented Kalman Filter. After that, you have to make semi-physical model of the building you will give these parameters to, so that the model can predict the future thermal behavior. Here, we do the same… except, we’re using data science tools instead of semi-physical model. By doing so, we get two advantages – much much less time needed to model everything and not having to dabble with complex mathematics. Interested how? See you at Tutorial 😉
Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
15:45

 —

17:15

Identification of Building Thermal Model – Using Data

Mihael Jaksic
Teaching and research assistant @ University of Zagreb Faculty of Electrical Engineering and Computing
Download

BREAKOUT SESSIONS

MR 24

MR 25 

MR 26

Breakout Break BS3

11:00

 —

11:30

Breakout Break BS3

Download

CO-LOCATED EVENTS

STREAM 4

Introduction to orchestration ETL processes with Apache Airflow

Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data, Data Engineering
Tool Showcase
Beginner to Intermediate
The aim of this tutorial is to demonstrate the usage of Apache Airflow to orchestrate ETL processes. Apache Airflow is an open-source workflow management platform for data engineering pipelines. In this tutorial a brief overview of architecture will be shown as well as core concepts of Apache Airflow such as DAGs, Tasks, Variables, Params and XComs. These concepts will be demonstrated through examples of ETL processes that are easily understandable and represent a good foundation to start to work with Apache Airflow. It is required to have some basic knowledge of Python programming language, running Docker containers and SQL language.
Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data, Data Engineering
09:00

 —

11:00

Introduction to orchestration ETL processes with Apache Airflow

Miroslav Tomic
Teaching Assistant @ University of Novi Sad, Faculty of Technical Sciences

Relation extraction from texts

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
Library Demonstration, Tool Showcase
Beginner to Intermediate
In this beginner-friendly Python tutorial, you’ll learn the basics of relation extraction using the OpenNRE toolkit. It covers the essential concepts and step-by-step instructions to extract relationships from text data without requiring extensive Python knowledge. By the end, you’ll understand the basic concepts of relation extraction so you can apply this valuable NLP technique to your own projects. Join us as we demystify the process of uncovering meaningful connections within unstructured text with ease!
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
11:15

 —

13:15

Relation extraction from texts

Andrija Poleksic
Assistant researcher @ Faculty of Informatics and Digital Technologies

From RNNs to Transformers: A Journey through Language Modeling

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation, Machine Learning, Deep Learning, GNNs, CNNs, RNNs
Career Path
Intermediate
This tutorial will follow the evolution steps of natural language processing and language models. We’ll begin with Recurrent Neural Networks (RNNs), explore how Attention Mechanisms revolutionized language modeling, and culminate with the architecture that shapes our present world – the Transformer. Through this journey we will also explore “How Small Can Language Models Be and Still Speak Coherent English?”.
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation, Machine Learning, Deep Learning, GNNs, CNNs, RNNs
15:45

 —

17:15

From RNNs to Transformers: A Journey through Language Modeling

Jovan Bubonja
Applied Scientist @ TomTom

Stream 5

Google Colaboratory and Python: Deep Learning basics for CV tasks

Image Classification, Image Processing, Machine Vision, Computer Vision
Career Path
Beginner to Intermediate
The Tech tutorial will cover the introduction to Google Colaboratory (e.g. how to enter Python code and run executable written commands) and some Python basics for images (e.g. how to load and display an image, and make changes on it). The last part of the tutorial will cover the basic concepts of Convolutional Neural Networks, and how to create, train and test a simple one.
Image Classification, Image Processing, Machine Vision, Computer Vision
09:00

 —

11:00

Google Colaboratory and Python: Deep Learning basics for CV tasks

Kristina Host
Research asistant @ Faculty of Informatics and Digital Technologies

Beginner’s guide to MRI brain scans classification using machine learning with FastSurfer, XGBoost and 3D CNNs

Image Classification, Image Processing, Machine Vision, Computer Vision, Machine Learning, Deep Learning, GNNs, CNNs, RNNs
Library Demonstration, Tool Showcase
Intermediate
This tutorial showcases two approaches to classifying medical data obtained through Magnetic Resonance Imaging (MRI) using machine learning techniques: Tabular volumetric data classification: The first method involves utilizing extracted tabular volumetric data. I’ll present FastSurfer pipeline and classification using XGBoost. Image Classification with 3D CNNs: The second method employs Convolutional Neural Networks to classify MRI scans as images. I will walk you through preprocessing steps, including skull stripping, scan registration, and intensity normalization using FastSurfer. Next, I will show how to create simple 3D CNN network in PyTorch.
Image Classification, Image Processing, Machine Vision, Computer Vision, Machine Learning, Deep Learning, GNNs, CNNs, RNNs
11:15

 —

13:15

Beginner’s guide to MRI brain scans classification using machine learning with FastSurfer, XGBoost and 3D CNNs

Damian Polak
Deep Learning Engineer @ Polish-Japanese Academy of Information Technology

Build Your Own AI Chatbot with OpenAI & LangChain

Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
Library Demonstration
Beginner to Intermediate
“Picture yourself crafting a chatbot with the power to engage, assist, and even entertain. Imagine the possibilities when you merge your coding skills with cutting-edge AI technology.

This is not just a dream; it can become your reality!

I am thrilled to invite you to this workshop titled ‘Build Your Own AI Chatbot with OpenAI & LangChain!’

During this 2-hour workshop, we will delve into the depths of conversational AI, exploring the transformative capabilities of OpenAI’s GPT models. I’ll guide you through the process of interacting with the OpenAI API, LangChain, handling API responses, managing conversation state, and so much more. Through real-world examples and interactive learning, we will empower you to create your own AI marvel.

This workshop is tailor-made for those with a programming background who crave the opportunity to leave their mark on the AI landscape. Whether you’re an aspiring developer, a curious tech enthusiast, or someone who simply wants to ride the AI wave, this workshop is for you. Together, we will unlock the full potential of AI and reshape the digital landscape.”
Natural Language Processing, Conversational AI, Speech Recognition, Machine Translation
15:45

 —

17:15

Build Your Own AI Chatbot with OpenAI & LangChain

Christophe Zoghbi
Founder & CEO @ Zaka

stream 6

Shinylive – Running R and Python Dashboards without a Server

Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
Tool Showcase
Intermediate
“R- and Python Shiny Dashboards typically need a server to run which can involve numerous infrastructure headaches for high availability and scaling. Thanks to the use of Web Assembly Shiny Dashboards can now also run directly in the browser without the need of complex server/cloud infrastructure.

In this tutorial we will show how to easily develop and deploy Shinylive dashboards and show the pros and cons of using this approach.”
Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
09:00

 —

11:00

Shinylive – Running R and Python Dashboards without a Server

Mario Annau
Managing Director @ Quantargo

Unleash the Crypto Data Landscape in your Hands

Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
Career Path
Intermediate
Explore the datasets we provide and build insights, models, visualizations, and results in order to win.
Data & Predictive Analytics, Data Visualisation, Time Series, Geospatial data
11:15

 —

13:15

Unleash the Crypto Data Landscape in your Hands

Maxim Legg
CTo and Founder @ Superchain
Download
Billetterie Weezevent