Munich Data Science #1

Thanks for your patience!

We are thrilled to finally announce our first meetup, which we organize together with the Munich Center for Machine Learning at LMU Munich. Come, grab a beer, meet your peers and learn about latest machine learning trends and cool applications of machine learning in data science. Our goal is a sound mix of talks which is of interest for both, folks from academia and industry.
For the first meetup we were lucky to win Vadim and Egor, who will talk about their own research:

Vadim Borisov: Dealing with categorical data in machine learning: von Null auf Held

Abstract:
Categorical features (variables) are quite common in many Data Science problems. However, it is more challenging to deal with them in comparison to numerical data. It can be explained by the fact that most of machine learning models require numbers as inputs; therefore, the categorical data must be encoded into numerical data.

In this talk, I will discuss the standard approaches for categorical data transformations as well as advance encoding approaches using the Gradient Boosting algorithm. Also, I will support my talk with practical code examples

Bio: Vadim is a Research Assistant at the University of Tübingen and part of the newly formed Data Science and Research (DSAR) group founded by Schufa Holding. His current research focuses on explainable and interpretable models in machine & deep learning. Out of school, Vadim regularly participates in machine learning competitions

Egor Labintsev: Text and image fusion for image retrieval and matching

Abstract:
Information about same entities in real world is often described by different modalities. For the humans it is common to perceive different modalities simultaneously, e.g. by looking at the product picture and reading its description. However, combination of information in different modalities is still a challenging task for machine learning algorithms.

In this talk, I will introduce several tasks and techniques in the area of text and image data fusion and discuss current research directions. I will explain cross attention and feature composition approaches in application to image retrieval and text-image matching

Bio: Egor is a machine learning researcher and a consultant in fields of NLP, recommender systems and time series prediction. His research interests include dialog systems and multimodal deep learning

Format:
• We start at 18:30
• 2 talks, 45 min each
• You can come hungry and thirsty, snacks and beer will await you :)
• We speak english