Data Science Seminars

The IT Academy Research Summit was a two-day event packed with highlight talks, poster sessions, free discussions, and speed-dating opportunities. You can watch here the recording of the plenary sessions.

• Recording of Day 1
  • Prof Jaak Vilo, Introduction, overview, and impact of the ITA programme
  • Assoc Prof Jaan Aru, Natural and Artificial Intelligence Lab
  • Assoc Prof Huber Flores, Distributed and Pervasive Systems Group
  • Assoc Prof Raivo Kolde, Health Informatics Group
  • Research Fellow Arnis Paršovs, Applied Cyber Security Group
• Recording of Day 2
  • ​​​​Assoc Prof Arun Kumar Singh, Collaborative Robotics and Robotic Computing Group
  • Assoc Prof Eduard Ševtšenko, Digital Supply Chain Group
  • Assoc Prof Naveed Muhammad,  Autonomous Driving Group
  • Mart Toots, Enterprise Estonia, formerly also ITL R&D Manager 
  • Dan Bogdanov, PhD, Cybernetica, Chief Scientific Officer, ITA Scientific Advisory Board member.
  • Prof Petteri Nurmi, University of Helsinki, ITA Scientific Advisory Board member.

The seminar was moderated by Prof. Jaak Vilo. The event featured the following speakers: 

• Liron Friend-Saadon, NVIDIA, Director of Developer Relations
• Kairit Sirts, University of Tartu, Associate Professor
• Zenodia Charpy, NVIDIA, Senior Deep Learning Data Scientist
• Fabian Weiss, NVIDIA, Senior Deep Learning SW Engineer
• Omar Ahmad, University of Iowa, Deputy Director, Driving Safety Research Institute
• Tambet Matiisen, University of Tartu Autonomous Driving Lab Lead Engineer

The seminar recording can be found here. The photos are available

The seminar was moderated by Marlon Dumas. The seminar featured the following speakers:

• Mohamad Gharib (UT) - Data/Information Quality Requirements– What is it and Why Does it Matter?
• Anastasija Nikiforova (UT) - Data lake or data warehouse? Data cleaning or data wrangling? How to ensure the quality of your data?
• Jürgen Koitsalu (Chief Product Owner, Swedbank Group Business Intelligence Product Management & Agile Guild)
• Igor Artemtšuk (Architect, Swedbank Group Business Intelligence Architecture Guild)

You can find the seminar recording here.

 The seminar featured the following speakers:

• Taavi Tillmann, University of Tartu, Integrating big health and social data, to better predict heart attacks, and reduce its socioeconomic inequality.
• Abdulrahman Azab, University of Oslo, Towards federation of cross-border sensitive data access and processing
• Liina Kamm, Cybernetica AS, Is usable privacy-preserving data analysis an oxymoron?
• Raivo Kolde, University of Tartu, Adopting common data model for Estonian health data
• Stefan Negru, CSC, Finland, Developing Technologies And Standards for Enabling Sensitive Data Archiving, Sharing and Reuse

You can find the seminar recording here.

The speaker Raivo Kolde has been supported by European Social Fund via „ICT programme“ measure.

Image

Seeing is believing and humans rely on their sight to make decisions. As a result, a lot of data we gather in various ways is visual. It is natural to want to get the most out of the data we have and one method toward achieving this is semantic segmentation. In essence, semantic segmentation means figuring out which pixels belong to which type of object. For example, how can a machine recognize a pedestrian on a road crossing or differentiate between fields and swamplands? Deep Learning has helped make huge strides in the field of semantic segmentation but there are still many challenges to overcome!

Speakers and presentations:

• Dmytro Fishman, University of Tartu, Lecturer of Artificial Intelligence - Biomedical segmentation: holy grail and remaining mysteries
• Tanya Shtym, KappaZeta, Machine Learning Engineer - KappaMask: AI-Based Cloud Mask Processor for Sentinel 2
• Markus Kängsepp, University of Tartu, Autonomous Driving Lab, Junior Research Fellow - Calibration of Bird's-Eye-View Semantic Segmentation
• Kaupo Voormansik, KappaZeta, CEO - Data Science and Machine Learning in KappaZeta

You can find the seminar recording here.

The benefits arising from Artificial Intelligence (AI) in terms of prediction accuracy, automation, new products and services or cost reduction are remarkable. But enterprises need to build trust and transparency in the data and algorithm used in AI systems to increase adoption. By default, AI systems such as machine learning or deep learning produce outputs with no explanation or context. As the predicted outcomes turn into recommendations, decisions or direct actions, humans tend to look for justification. Most experts in the field agree that AI systems should be less ambiguous to the end-users and subjects of algorithmic decision-making. In this seminar, four speakers from the University of Tartu and Wise will discuss the explainability and transparency of AI as well as the requirements and challenges of building a robust machine learning system.

Speakers and presentations:

• Meelis Kull, Associate Professor in Machine Learning at UniTartuCS: Understanding the decisions of an AI system
• Radwa El Shawi, Associate Professor of Big Data at UniTartuCS: Towards automatic concept-based explanations
• Luca Traverso, Data Science Lead at Wise: Building robust and defensible machine learning systems in Fincrime
• Rain Vagel, Data Scientist at Wise: Why did the model decide X?

You can find the seminar recording here.

Moderators:

• Ahmed Awad, Professor of Data Systems @ University of Tartu, Estonia
• Feras M. Awaysheh, Assistant Professor of Data Analytics @ University of Tartu, Estonia

Speakers:

• Andreas Hellander (University of Uppsala): Scalable Federated Machine Learning with FEDn
• Peter Richtarik (KAUST): EF21: A new, simpler, theoretically better, and practically faster error feedback
• Aaron Ardiri (RIoT Secure): The Internet of Disconnected Things
• Essam Mansour (Concordia University): A Data Discovery Platform Empowered by Knowledge Graph Technologies: Challenges and Opportunities

You can find the seminar recording here.

Moderator: Mark Fišel (UniTartuCS)

Speakers and presentations:

• Anna Mischenko (AI_NORN): AI_NORN – programming robots as an art
• Eduard Barbu (UniTartuCS): Devices to compute human creativity
• Anna Aljanaki (UniTartuCS): Let there be music: when AI learns to compose
• Jaanus Jaggo (UniTartuCS): The secret of making an endless world for a video game

You can find the seminar recording here.

Moderator: Riccardo Tommasini, Lecturer of Data Management at UniTartuCS

Speakers and presentations:

• Emanuele Della Valle (Politecnico di Milano; Quantia Consulting): Taming velocity: A tale of four streams
• Matteo Ferroni (Bottega52): Fight data gravity with event-driven architectures
• Niccolò Raspa (SIGHUP): Accelerating data science with MLOps and cloud native technologies

You can find the seminar recording here.

Moderator: Ülar Allas from High Performance Computing Center of UniTartu

Speakers and presentations:

• Ott Oopkaup (UniTartu, High Performance Computing Center): High performance computing: A look into how large-scale data storage and processing is done
• Reedik Mägi (UniTartu, Estonian Genome Centre): Computational challenges in genomics: association studies, genetic risk scores and AI
• Kaur Alasoo (UniTartu, Institute of Computer Science): Automating the processing of terabyte-scale data with software containers and workflow engines
• Velle Toll (UniTartu, Institute of Physics): HPC for weather, climate and Earth system prediction
• Pekka Manninen (LUMI supercomputing facility): LUMI: the European flagship supercomputer of the North

You can find the seminar recording here.

Moderator: Elena Sügis (UniTartuCS)

Speakers and presentations:

• Markus Lippus (MindTitan): Delivering on 85% of your AI projects
• Kalev Koppel (STACC): Hired or fired by AI
• Marlon Dumas (UniTartuCS): Process mining in action
• Jaak Vilo (UniTartuCS): Supply chain for data scientists
• Elena Sügis (UniTartuCS): Data science 101 for your business

You can find the seminar recording here.

(Online)

Introduction to the topic by:
Kaur Alasoo & Riccardo Tommasini (University of Tartu Institute of Computer Science)

Speakers and presentations:

• Łukasz Grądzki (Bolt): Data Platform at Bolt: Lessons from scaling data infrastructure in a hyper growth company
• Kristjan Eljand (Eesti Energia): Labelling the labelled
• Rao Pärnpuu (Starship Technologies): Using datasets to develop and globally operate self-driving robots
• Taivo Pungas: Datasets: the source code of Software 2.0.

You can find the seminar recording here.

(Online)

Moderator: Jaak Vilo (UniTartu ICS)

Speakers and presentations:

• Krista Fischer (UniTartu, Estonian Genome Center): Nowcasting and forecasting of COVID-19 in Estonia: experiences from spring 2020
• Hedi Peterson (UniTartu ICS): COVID-19 and us. Let the data speak.
• Raivo Kolde (UniTartu ICS): Creating (inter)national COVID-19 evidence base through health data standardisation
• Dan Bogdanov (Cybernetica, HOIA.me): COVID-19 contact tracing apps in Estonia and abroad

You can find the seminar recording here.

Moderator: Tambet Matiisen (UniTartu Autonomous Driving Lab)

Speakers and presentations:

• Naveed Muhammad (UniTartu ICS): Autonomous driving – past, present and future
• Martin Appo (Cleveron): Cleveron’s journey towards driverless delivery vehicle
• Sergey Kharagorgiev (Starship Technologies): Computer vision for obstacle avoidance in the wild
• Alex Kendall (Wayve): Creating an artificial driving intelligence

Moderator: Alexander Nolte (UniTartu ICS)

Speakers and presentations:

• Marlon Dumas (UniTartu ICS): Data science and AI for business process improvement
• Markus Lippus (MindTitan): The unexpected use cases for a machine that can listen
• Sven Laur (STACC): Health insurance analytics: A case study at the Estonian Health Insurance Fund
• Lauri Antalainen (CoreGrow): Optimizing production processes: how can data science help?

You can find the seminar recording here.

Moderator: Meelis Kull (UniTartu ICS)

Speakers and presentations:

• Lauri Sokk (Smart City Tartu): Smart City Tartu – why we do what we do?
• Hans Leis (Bercman Technologies): The Smart Pedestrian Crosswalk
• Anti Gruno (Datel): Metallica concert through SAR eye, using Datel’s Early Warning system SILLE
• Roman Meeksa (Tartu City Government): Tartu Smart Bike Share – how and what do we see?

You can find the seminar recording here.

Moderator: Meelis Kull (UniTartu ICS)

Speakers and presentations:

• Ando Saabas (Taxify/Bolt): Interpreting tree-based models
• Anna Aljanaki (Mooncascade): What music information retrieval can tell us about Eurovision?
• Kairit Sirts (UniTartu): Understanding neural models for text analysis
• Markus Lippus (MindTitan): Trust the machine, or do you really need to know what your algorithm is doing?

You can find the seminar recording here.

Moderator: Sherif Sakr (UniTartu ICS)

Speakers and presentations:

• Mohamed Maher (UniTartu): SmartML – Towards Optimized Automated Machine Learning Pipelines in the Big Data Era
• Felix Mohr (Paderborn University): ML-Plan – Automated machine learning via hierarchical planning
• Martin Strohbach: IoTCrawler – Building a Search Engine for the Internet of Things
• Mihkel Solvak (UniTartu): Anonymized i-voting log data: how can it be used or abused to understand voter behavior?
• Lauri Sokk (Tartu City Government): Smart City since 1632

You can find the seminar recording here.

Moderator: Amnir Hadachi (UniTartu ICS)

Speakers and presentations:

• Margus Tiru (Positium): Mobile Positioning Data for Human Spatio-Temporal Behavioural Analysis
• Toivo Vajakas (Reach-U): Some thoughts on making use of passive mobile positioning data
• Kalev Koppel (KappaZeta): Deep learning for satellites based grasslands monitoring. Lessons learned
• Mikhail Iljin (Taxify): Real-time rebalancing of demand and supply at Taxify

You can find the seminar recording here. 

Moderator: Leopold Parts (UniTartu ICS)

Speakers and presentations:

• Krista Fischer (UniTartu, Estonian Genome Center): Disentangling causes and effects – how can genetics help?
• Taavi Tamkivi (dataminer.ee): Being a good cop of the data world – how to find criminals behind the lakes of data
• Maris Alver (UniTartu, Estonian Genome Center): Implications of big data for clinical management of cardiovascular disease
• Andrei Tsõmbaljuk (TransferWise): Machine Learning at TransferWise
• Andres Võrk (Tni Tartu, CITIS): Examples of data-driven policy impact evaluation in Estonia

You can find the seminar recording here.

Moderator: Mark Fišel (UniTartu ICS)

Speakers and presentations:

• Tanel Alumäe (TalTech): Speech Recognition
• Silver Traat (TEXTA): Use cases from TEXTA
• Sven Laur (STACC, UniTartu ICS): EstNLTK libraries for NLP
• Risto Hinno (FeelingStream): Daily challenges with text mining
• Kairit Sirts (UniTartu ICS): Clinical text Classification

You can find the seminar recording here. 

Moderator: Leopold Parts (UniTartu ICS)

Speakers and presentations:

• Marek Rei (University of Cambridge): Human Interpretability of Machine Learning Models
• Mihkel Solvak (UniTartu): Real-Time Predictive Economics
• Toomas Kirt (Statistics Estonia): Big Data in Statistics
• Kaspar Märtens (Oxford University): Modern Frameworks for Automated Inference
• Krista Fischer (UniTartu): Disentangling Correlation and Causality in the Big Data Era
• Alex Graves (DeepMind): Associative Compression Networks For Representation Learning

You can find the seminar recording here. 

Moderator: Tambet Matiisen (UniTartu ICS)

Speakers and presentations:

• Raivo Sell (TUT): ISEAUTO – the first Estonian self-driving car project with Silberauto
• Lindsay Roberts (Starship Technologies): Visual Localisation in Urban Environments
• Lauri Tammeveski (Milrem Robotics): How to recognize different types of trees from quite a long way away
• Allan Aksiim (Foundation for Future Technologies): Laws, Regulations and Ignorance
• Andrej Karpathy (Tesla): The Challenges of Applying Machine Learning for Autonomous Vehicles (video broadcast)

You can find the seminar recording here. 

Moderator: Dmytro Fishman (UniTartu ICS)

Speakers and presentations:

• Kaupo Palo (PerkinElmer): Microcopy image analysis
• Karl Kruusamäe (UniTartu): Opening Machine's Eyes: Why We Need Image Analysis on Robotics?
• Kaupo Voormansik (KappaZeta): Satellite Imagery Time Series Processing
• Gholamreza Abarjafari (UniTartu): DeepVision for Human Behaviour Analysis

You can find the seminar recording here. 

Moderator: Raul Vicente (UniTartu ICS)

Speakers and presentations:

• Jaan Aru (UniTartu ICS): Deep Learning and the Brain 
• Robert Roosalu (MindTitan): Enterprise Deep Learning
• Tambet Matiisen (UniTartu ICS): Keras Advanced Tips & Tricks
• Hendrik Luuk (AlphaBlues): Automating customer service chat with AI

You can find the seminar recording here. 

Moderator: Meelis Kull (UniTartu ICS)

Speakers and presentations:

• Kaur Alasoo (UniTartu ICS): Uncovering Hidden Biases in Data with Visualisation
• Tormi Reinson (Pipedrive): Communicating Data with D3.js
• Anto Aasa (UniTartu): Visualizing Data: Space & Time
• Mirko Känd (IxD.ma): Everyone is a Data Scientist

You can find the seminar recording here.

Moderator: Leopold Parts (UniTartu ICS)

Speakers and presentations:

• Ilya Kuzovkin (UniTartu ICS): Deep Learning Zoo
• Mark Fišel (UniTartu ICS): Neural Machine Translation
• Tanel Pärnamaa (Fits.me): A Neural Knowledge Language Model
• Tambet Matiisen (UniTartu ICS): Life in OpenAI

Moderator: Leopold Parts (UniTartu ICS)

Speakers and presentations:

• Jaak Vilo (UniTartu ICS): Data for Health
• Eneli Oitmaa (Asper Biotech): Applying Bioinformatics Analysis in the Genetic Laboratory Settings
• Dmytro Fishman (UniTartu ICS): Deep Learning in Health Care
• Lili Milani (Estonian Genome Centre): Electronic Health Records and Genomes for Research

You can find the seminar recording here. 

Moderator: Kristjan Eljand (STACC)

Speakers and presentations:

• Fredrik Milani (UniTartu ICS): Business Process Mining
• Martin Märss (Swedbank): BI Concept in a Bank
• Marlon Dumas (UniTartu ICS): Predictive Business Process Monitoring
• Ester Eggert (TransferWise): Applying ML to Create Money without Borders
• Nicola Vitucci (Open Data Day): Linked Data: A Quick Introduction

You can find the seminar recording here. 

Moderator: Leopold Parts (UniTartu ICS)

Speakers and presentations:

• Leopold Parts (UniTartu ICS): Introduction
• Raul Vicente (UniTartu ICS): Machine Learning in the Lab
• André Karpištšenko (Proekspert, PlanetOS): Machine Learning in Production
• Vladimir Šor (Plumbr): Java Performance Monitoring at Plumbr
• Mark Fišel (UniTartu ICS): Natural language Processing
• Taavi Tammiste (Mooncascade): Finding Real World Use-cases for Data Science
• Amnir Hadachi (UniTartu ICS): Urban Sensing: From Mobility to Localization