This is a special live telecast of first ever remote Berlin Machine Learning Meetup. Visit us at:
------------------------------------------------
Talk 1: How to classify news articles in the “real world”?
Speaker: Helena Amankaya Peña (Upday) and Malgorzata Adamczyk (AS-Ideas)
Abstract: Every day at Upday we serve over 85K news articles to millions of users across Europe. This means we process a lot of textual data in many languages and contexts. In order to connect people with the right content we need to know what the articles are about - we need to classify them. The goal of our latest project was to replace a rule-based classification system with an ML model. Our new model should be fast, easy to scale to further markets and perform well with little training data. In this talk, we will present our journey to finding an algorithm that meets these expectations.
Bios:
-Helena is a data scientist at the digital news company Upday. Her drive is to work towards useful, human-friendly applications of data! She studied biomathematics and completed a PhD in mathematics at the University of Greifswald, Germany.
-Malgorzata works as an ML at Axel Springer AI, the Artificial Intelligence unit of Axel Springer SE. Previously she was a member of Data Science team at idealo.de. She worked on projects in domains of NLP, Computer Vision and currently is exploring the potential of GANs for image generation.
--
Talk 2: imagededup - Finding duplicate images made easy
Speaker: Tanuj Jain (Axel Springer AI)
Abstract: The problem of finding duplicates in an image collection is widespread. Many online businesses rely on image galleries to deliver a good customer experience and consequently, generate more revenue. Hence, the image galleries need to be of the highest quality. Duplicates in such galleries could degrade the customer experience. Additionally, duplicates can lead to wrong evaluation of image-based ML models which would degrade its true performance. Therefore, finding and removing duplicates is of high importance. In this talk, we want to present imagededup, a Python package that we built to solve that specific problem of finding exact and near duplicates in an image collection. We will speak about the motivation why we need to build it, its functionality and also give a demo.
Bio: I am working as a Senior ML Engineer at Axel Springer AI since December 2019. I was previously part of the Data Science team at idealo Internet GmbH. My current interests revolve around deep learning research for speech and image processing. I completed my M.Sc. in Electrical Engineering from Paderborn University in 2015 and have been working in the field of Data Science ever since. I’m interested in leveraging the power of machine learning to empower businesses and measure the impact thus created.
--
Talk 3: Use cases of explainable AI in Python
Speaker: Karol Przystalksi (Codete)
Abstract: In the days where we have autonomous cars, drones, and automated medical diagnostics, we'll want to learn more about how to interpret the decisions made by the ML models. This information helps debugging and more efficient model retraining. This talk explains the taxonomy of explainable models, and approaches to explainable AI (XAI). We go through regression methods, decision trees, ensemble methods, and neural networks.
Bio: Karol Przystalski is CTO and founder of Codete. He obtained a PhD in Computer Science from the Institute of Fundamental Technological Research, Polish Academy of Sciences, and was a research assistant at Jagiellonian University in Cracow. His role at Codete is focused on leading and mentoring teams. The company has built a research lab that is working on ML methods and big data solutions in specialty areas such as pattern recognition and HDP implementation.
-----------------------------------------
Follow me on:
Twitter:
LinkedIn:
Kaggle:
0 Comments