14, 15, 16 November 2019

9:00 - 10:00
Registration / Networking

10:00 - 10:20
Main Stage
10:20 - 10:40
Main Stage
Designing for impact: the power and potential of the motley crew
Bronwen Robertson
Researcher, Data4Change
DATA4CHANGE connects Civil Society Organisations with global and local experts in data, design, tech and journalism to create powerful data-driven advocacy projects that forge real change. Want to know our trade secrets? Through case studies from our flagship sprints, Bronwen will show you how to unlock the power and potential of the motley crew through curating transformational collaborations.
10:40 - 11:00
Main Stage
Data 4 Impact
Ruzanna Baldryan
Design Thinking & Communications Lead, National SDG Innovation Lab
Data is the backbone of decision-making: the data drawn from mobile communications, remote sensing, and social media is used by the private sector for marketing, advertising, and management. The same data holds valuable insights for policy and the public good. During the talk, Ruzanna will cover the ways data science can discover new realms in the use of data for sustainable development challenges ranging from people's wellbeing to climate change.
10:00 - 11:00
The EU for Civil Society Activists Room
Drawing the Maps: Gerrymandering and Redistricting
Mike Sager
Chief Technology Officer, EMILY’s List
In the United States, political district maps are often drawn by the parties in power, which is then used to further consolidate that power. EMILY's List Chief Technology Officer Mike Sager will walk through different examples of gerrymandered districts from the United States, and then lead participants through a "Redistricting Game" where they draw their own maps that take into account all the applicable political considerations.
Participants should have laptops.
10:00 - 11:00
ODIHR Human Rights Room
Story about Femicide in the Conflict Zone - from project idea to a large social campaign
Tina Tsomaia & Ia Shalamberidze
Professor, Georgian Institute of Public Affairs
Deliberate murder of women in an armed conflict is one of the forms of femicide, according to The Vienna Declaration (2012). Tiko Tsomaia and Ia Shalamberidze will present their transmedia project that tells the stories of murdering of 13 women during the war in Abkhazia, Georgia. The aim of the workshop is share the experience from idea initiation to turning it into a platform of various uses from therapy to a big source of information.
You'll need to have a laptop to participate in this workshop.
10:00 - 11:00
ForSet DataViz Room
Data Looks Better Naked (Part I)
Joey Cherdarchuk
Co-founder and Design Lead, Darkhorse Analytics
Too often when we visualize our data we try to dress it up when we should be stripping it down. In this workshop Joey will walk you through some of the common pitfalls people encounter when trying to communicate with data. He'll review best practices to ensure you understand your data and that your data speaks to your audience.
10:00 - 11:00
BoG Fintech Room
Building Data Science Organization with Business Impact in Mind
Mikheil Nadareishvili, Mariam Lelashvili & Levan Borchkhadze
Deputy Head of BI, TBC Bank
There are many challenges associated with building a data organization at an established company. One of the most important and interesting ones is how to make sure that data science efforts actually solve business problems and translate into results. The workshop will focus on how TBC Bank addressed this problem.
10:00 - 11:00
Maxin AI Room
How to beat the Kaggle benchmark
Andjela Todorovic
Software Engineer, Cubic Corporation
Welcome to the world of data science competitions! We will practice the data preprocessing, analyzing and EDA in the first step, then train different models and comparing them in scores. Following, we will discuss some of the new boosting techniques such as XGBoost, AdaBoost etc. that have shown great results on Kaggle leaderboards.
You'll need to have a workshop at this workshop.
11:00 - 11:20
Main Stage
How data-driven storytelling can amplify social impact
Jacopo Ottaviani
Head of Data, Code for Africa
Data-driven storytelling has become a notorious entity in the media landscape and in the global civil society. As a multi-disciplinary field that combines, among others, computer science, information design and journalism, it is hard to provide a fully comprehensive definition, but it's proven a key tool to amplify social impact around the world.

11:00 - 11:20

11:20 - 11:50
Main Stage
On large scale, data driven human rights investigations
Milena Marin
Senior Advisor for Tactical Research, Amnesty International
The landscape for human rights investigators has changed dramatically over the last 10 years. Conflict zones and other human rights hot spots are emanating unprecedented levels of digital data-streams. Human rights investigators need to expand their tool kit beyond field missions and direct testimonies to take advantage of the wealth of information at their fingertips. In this session, Milena will discuss how Amnesty International is adapting to this fast-paced environment combining field missions with open source investigations, crowdsourcing, and data science to expose human rights violations at scale.
11:50 - 12:20
Main Stage
Data Walking
David Hunter
Associate Senior Lecturer, Ravensbourne
Data Walking is a research project exploring the potential of walking to gather environmental data and through multiple walks and visualisations build a rich picture of that area. The project examines technology and tools for creative data gathering and experimenting with data visualisation, to create tools, gain insight, and share knowledge. In this talk, David will look back discussing and displaying the output of the last five years of the project and look forward to future goals and ambitions.
11:20 - 12:20
The EU for Civil Society Activists Room
Cybersecurity for everyone: How to ispire yourself and your friends around to protect your Data!
Mikalai Kvantaliani
Director, Association New Group
During the workshop, attendees will learn what are the trends, threats & solutions for data protection; why it is important for yourself, your colleagues and your country to know basics of cyber security for everyone; who is your personal adversary, and what are the tools and approaches to protect yourself; who are the main stakeholders in personal data protection. Mikalai will also go through the solutions & tools we can use to protect our data.
11:30 - 12:30
ODIHR Human Rights Room
Machine learning concepts for human rights activists
Chris Kennedy
Data Scientist, Berkeley Institute for Data Science
This workshop will give a high-level background on machine learning, with a special focus on human rights application and anti-Semitism. What is machine learning? What can it do? What kinds of data can it analyze? What is the difference between deep learning and machine learning? On the application side, the workshop will explore use cases for machine learning to address important issues across human rights, especially with text and image-related data.
11:30 - 12:20
ForSet DataViz Room
Data Looks Better Naked (Part II)
Joey Cherdarchuk
Co-founder and Design Lead, Darkhorse Analytics
Too often when we visualize our data we try to dress it up when we should be stripping it down. In this workshop Joey will walk you through some of the common pitfalls people encounter when trying to communicate with data. He'll review best practices to ensure you understand your data and that your data speaks to your audience.
11:20 - 12:20
BoG Fintech Room
Integrating international FinTech solutions into Georgian system (Bank of Georgia case study)
Stefi Gual & Davit Lapiashvili
Product Manager for Digital Channels at Strands, Strands & Solution Architect, Bank of Georgia
11:20 - 12:20
Maxin AI Room
Computer shows why - visualization methods for interpretting and dubugging deep learning models
Eyal Gruss
Independent consultant
Deep neural networks are the modern day Swiss-knife for image and text processing, understanding and generation. Eyal will review the toolbox of visualization methods for evaluating, debugging and interpreting the performance and decisions of neural networks and machine learning algorithms. He will present some real-world usages that he applied in the medical domain. You will do hands-on experimentation with salient heatmap techniques for images and videos, and will get the know-how and code templates for applying these methods to your own models and data.
participants should have laptops. python. hands-on familiarity with convolutional neural networks.

12:20 - 14:00
Networking Break

14:00 - 14:20
Main Stage
Design Decisions in Data Visualization
Harry Stevens
Visual Journalist, Washington Post
Designing a data visualization requires a series of decisions, from the large to the small. What visualize representations best communicate your data? What colors should you use? Where should you place the labels and annotations? This talk will present techniques for making good decisions in data visualization design.
14:20 - 14:40
Main Stage
We are data people: Driving success with data-driven culture
Beso Elbakidze
Visual Analytics, Deloitte
How does your organisation make the pivot to becoming insight driven, in day-to-day working, core business processes and most importantly, in decision-making?
14:40 - 15:00
Main Stage
Getting started with data driven product decisions
Danielle Diaz Dussan
Data Scientist,
Does your product generate data? Then you should use it! This talk will give you simple examples of how Data Scientists at work together with product teams to help them make better decisions, followed with advice on how to start doing it too.
15:00 - 15:20
Main Stage
Data for Business
Alexander Filatov
Head of Analytics, Visa CIS
Alexander will explain how most companies have built their internal AI and ML expertise, also what are the main issues they have faced and what are the ways to avoid such issues in your teams.
14:00 - 15:20
The EU for Civil Society Activists Room
How to build your own project without big fuckups
Jane Klepa
Executive Director, 1991 Open Data Incubator
Practical aspects in developing project. Milestones, technics, do's and don'ts
14:00 - 15:20
ODIHR Human Rights Room
Let's spend a bit more time on data security than on coffee!
Giorgi Iashvili
Information Security Consultant
14:00 - 15:20
Forset dataviz room
Introduction to Shiny: building interactive dashboards with R
Ildiko Czeller
Lead Data Scientist, Emarsys
The {shiny} R package enables you to build interactive dashboards and web applications without requiring javascript or html knowledge. An interactive web application can help you immensely during data analysis and is also a great way to showcase your results to non-technical stakeholders. During the workshop we will walk through the most important concepts and building blocks while creating a realistic example shiny application.
For this workshop you'll need: a laptop and beginner experience in R: ability to create plots and summary tables in R.
14:00 - 15:20
BoG Fintech Room
Forecasting time series data with recurrent neural networks
Edgar Khachatryan
Quality Assurance Engineer, Questrade Inc.
A time series is a type of numerical sequential data ordered by time. Traditionally, statistical methods can perform the analysis and prediction of such data, however, these models require frequent updates to match the continuously changing patterns over time. Modern algorithms and architectures in the machine learning field are making great strides in the prediction of sequential data, including time series during recent years. The workshop will illustrate some common steps and outline mistakes that are made during time series analysis using Deep Neural Networks.
General understanding of Machine Learning is a requirement for this workshop.
14:00 - 15:20
Maxin AI Room
Solving real-world decision making problems with constraint programming
Dr. Christian Drescher
Member of Artificial Intelligence Research group, Mercedes-Benz
Industry is increasingly looking at automating cognitive tasks, including decision making in challenges from operations research like production sequencing, staff rostering, job scheduling, and bin packing. When using computers to automate such combinatorial problems, it is often easier to declare what a computer program must accomplish rather than describing how to accomplish it as a sequence of instructions. Constraint Programming follows the declarative programming paradigm by allowing users to state the properties of the feasible solutions for a set of decision variables, and draws on a wide range of constraint solving techniques from Artificial intelligence to automatically determine solutions to any instance of the stated problem. This workshop gives a brief introduction to Constraint Programming using examples that are accessible to a general audience, like solving Sudoku puzzles. Part of the session allows for hands-on activity using the open source Constraint Programming toolkit MiniZinc. Participants are encouraged to have the MiniZinc IDE installed on their laptop if they wish to experience Constraint Programming in practice. However, it is not required to follow the contents of the workshop.
MiniZinc IDE

15:20 - 16:00

16:00 - 16:20
Main Stage
Post-Mortem: How We Failed to Apply Machine Learning to International Trade Regulation
Anton Prokopyev
Former Data Scientist, The World Bank
The World Bank and the United Nations Conference on Trade and Development (UNCTAD) aim to remedy gaps in the collection and analysis of data related to international trade regulation. To this date, much of this work is being done manually by consultants reading thousands of pages of government documents. This talk will cover the efforts made to automate the identification of government documents that put a restriction on international trade, and why this was unsuccessful.
16:20 - 16:40
Main Stage
Engaged and outraged
Zdenek Hynek
Developer, Data4Change
Are we in control of how we spend our free time? With sophisticated AI systems designed to keep us hooked, do we even have a choice to choose what we watch? This talk explores the impacts of the Youtube recommendation algorithm on politics, society and culture.
16:40 - 17:00
Main Stage
Machine Learning - The Good, the Bad and the Ugly
Benjamin Auffarth
Lead Data Scientist, Oleeo
In this talk, Ben will discuss common challenges of machine learning projects, do's and don'ts, and he will go through best practice recommendations for successful projects.
17:00 - 17:40
Main Stage
How to mute background noise with AI
Anush Martirosyan
Project Manager, Krisp
The workshop will explore the specifics of audio signals and the methods of extracting important features from the signal. Anush will also talk about some deep learning techniques that her companu has used to build a real-time working Noise Cancellation algorithm - starting from the data collection to the inference and evaluation. You will learn about latency constraints and the tradeoff between the quality and the processing time.
16:00 - 17:40
The EU for Civil Society Activists Room
Investigating inauthentic activities on Facebook and Twitter
Eto Buziashvili
Researcher, Atlantic Council’s Digital Forensic Research Lab
Are you on Facebook or Twitter? Then we should talk about coordinated inauthentic behavior and bots. What actually are they and how to spot them? Learn how to investigate inauthentic activities and join the #DigitalSherloks movement!
You'll need to have a laptop to participate in this workshop.
16:00 - 17:40
ODIHR Human Rights Room
Visualization of asset and income declaration of Gerogian public officials: pandas and datawrapper
Andrew Jvirblis & Artem Shchennikov
Head, & Developer, Transparency International Russia
The workshop will provide the participants with a data sample parsed from official declarations of Georgian public officials for 2018. The attandees will learn how to narrow the dataset and build a csv from the provided json using pandas and how to use the received file in the DataWrapper service to build a map showing the preference of Georgian politicians in realestate, differed by location, type, size. We will show one of the possible tracks for use of data, while the provided data sample can be visualized in another way at the choice of participants.
Participants should have some experience in python, had it installed on their laptops together with pandas library. An experience working with json data is welcome but not required.
16:00 - 17:00
ForSet DataViz Room
Hands-on Visualization of Geographic Movement: How to make an interactive flow map
Ilya Boyandin
Data Visualization Engineer, Teralytics
Flow maps are one of the most widely used visualizations of geographic movement. They represent aggregated numbers of movements between locations (e.g. numbers of migrants or the amount of goods transported) as lines of varying thicknesses on a map. Until recently making flow maps involved either manual drawing, or knowledge of programming, or specialized software. There hadn't been easy to use tools with which people could make flow maps and share them online. With, a free and open-source web application Ilya and his colleagues developed, anyone can create a flow map from data in a Google spreadsheet in no time. Since the tool has been released, people from different parts of the world started using it to visualize various mobility datasets. At the workshop Ilya will demonstrate the tool, show example datasets and discuss a few interesting patterns which can be discovered in them. He will also touch on the analysis of changes over time, on representing flows with attributes and on the scalability of the general approach to very large datasets. Bring your laptop: you will create at least one flow map.
Don't forget a laptop if you want to attend this workshop.
16:00 - 17:00
BoG Fintech Room
Rage Against The Machine: Fighting Bias in Data
Sophie Shawdon
Senior Data Analyst, ClearScore
Companies around the world are becoming more data-driven in order to be more objective, and less reliant on gut instinct and human bias. But what happens when the data or the models themselves become biased? In this workshop, Sophie will discuss real life examples of bias from different industries; talk through common traps and how they occur; and give you the tools you need to fight them.
16:00 - 17:00
Maxin AI Room
Automating Customer Support in a large organization
Markus Lippus
Co-founder & Data Science Lead, MindTitan
The existence of RPA beautifully illustrates that automating things in a large organization is way complicated. Using AI and machine learning only further complicates the adoption process. In this workshop Markus'll go over the problems a customer service department in a large company faces and the ways machine learning is being used to help. He will talk about the different solutions and their pros and cons. And most importantly he'll talk about the organizational change and adaptation that making a meaningful impact requires and the kinds of opportunities it opens.
17:40 - 18:00
Main Stage
18:00 - 20:00