PyData Warsaw #22

PyData Warsaw is back!

We have a few extremely interesting talks for you! Thanks to our friends at Allegro we will be able to meet in a great and comfortable venue and listen to great Speakers.

18.00 - Piotr Wygocki - CEO at MIM Solutions - "FOLLISCAN: how to automate an ultrasound examination with deep learning?"

Abstract: Ultrasound is one of the most common medical imaging techniques. FOLLISCAN is an analytical and predictive system based on deep learning algorithms that supports doctors in ultrasound diagnostics of the ovaries - examination of antral follicles. Follicle counting is of great practical importance, e.g. it helps to estimate ovarian reserve. In addition, a large number of antral follicles indicate polycystic ovarian morphology. At the meeting, I will talk about the technological challenges we encountered while creating the solution and how we solved them.

18.45 - Piotr Migdał - Head of AI @ Dali Games - "Limitations of content generation with AI (spoiler: the sky's the limit)"

Abstract: Generative models, such as ChatGPT and Stable Diffusion, started with a bang and now get better every few months.
I will share a few creative uses, mention their limitations... and explain how we can overcome these limitations.
Also, there will be a simple on what we do at Dali Games - end-to-end mobile game generation, from a single prompt.

19.30 - Philipp Peterseil - PhD candidate and university assistant @
Institute for Communications Engineering and RF-Systems
Johannes Kepler University Linz - "Trustworthiness in UWB Ranging and Localization"

Abstract: The Internet of Things (IoT) is a network of billions of connected hardware-constrained devices, which enable smart scenarios in different contexts, e.g. smart factories. These connected devices can react to the environment and receive services according to their locations. A widespread state-of-the-art option to obtain accurate range-based location information is ultra-wideband (UWB) wireless communication. Thereby, trustworthiness is seen as a promising way to ensure the reliability, safety and security of such IoT systems. To evaluate a trustworthiness score for ultra-wideband distance estimations, we came up with two different machine learning based approaches: a joint estimation of error correction and trust based on a modified version of K-nearest neighbours (KNN) and a trustworthiness indicator based on long short-term memory (LSTM) based autoencoder. In the talk, both structures will be presented and their performance will be evaluated.

Venue:
We are meeting at Allegro HQ in Warsaw:

Norblin Factory (Allegro Office - Plater building, entrance Plater 3 from Żelazna street)
ul. Żelazna 51/53,
00-841 Warsaw

Please remember to use Plater 3 entrance from Zelazna - we will meet you there!

If you cannot make it in person - we have Zoom provided: https://allegro.zoom.us/j/95922763768?pwd=bVpZMFVvTkZyZ213ZkZMYys0VzN1dz09

Information about car park and bike park are available here: https://fabrykanorblina.pl/en/location/

Public transport
Directly next to one of the historic buildings of the Norblin Factory is the Norblin 05 stop served by 109, 178, N45 and N95 buses.
From Żelazna Street, Norblin 06 stop is served by the 157 bus.

Not far away are two M2 metro stations, Rondo Daszyńskiego and Rondo ONZ.

Buses serving Rondo ONZ: 109, 160, 174, N12, N13, N14, N16, N21, N44, N46, N62, N63, N64, N71, N95

Trams serving Rondo Daszyńskiego: 1, 9, 11, 14, 22, 24, 25