Codete Meetup #1

AGENDA:

1️⃣ Karol Przystalski - Buzzword-driven Machine learning

We go through the most popular mistakes done in ML/DS projects. Show how to avoid it and build a successful AI project. This presentation is dedicated to managers and decision-makers in the software and data science departments.

2️⃣ Kacper Łukawski - Autoencoders for anomaly detection

During the talk, we are going to introduce an idea of the autoencoder and present some common use cases, with a strong emphasis on anomaly detection. This method allows utilizing unlabelled datasets in an unsupervised way and is surprisingly well adapted in Computer Vision and Natural Language Processing tasks. We will review some examples of applying autoencoders in these areas as well.

3️⃣ Piotr Wawryka - How to fool neural networks

Neural networks are commonly used machine learning models that have achieved state of the art performance on a variety of tasks, e.g., image or speech recognition. Although they seem to be a powerful tool, neural networks are extremely susceptible to a specific kind of attack. In my presentation, I would like to present to an audience what are so-called adversarial examples that lead neural networks to misclassification. How to create such examples. And finally, I would like to show how to defend our model against those kinds of attacks.