Past Events

  • Gradient Boosting Workshop - Anand Subramoney & Adrian Spataru

    Gradient Boosting is a machine learning algorithm used mostly for regression and classification task. Such Methods have been used successfully to win machine learning competitions on Kaggle. Furthermore, gradient boosting is being more and more used in the industry. In this workshop, we will go first through the theory to get a better intuition and understanding. Finally, we will analyze different case studies and provide practical tips on using gradient boosting in your projects.
    The Workshop is split into 2 parts:
    1. Part - Anand Subramoney will give a theoretical overview of Gradient Boosting.
    2. Part - Adrian Spataru will present how you can apply different implementations of the algorithm and present several case studies.

    Slides Part1+Part2
  • Dealing with Imbalanced Data - Adrian Spataru

    Imbalanced data is a common problem in lots of domains, such as Fraud detection, Disease detection etc. In this talk, we will cover different methods for dealing with such problems. For example: SMOTE, ADASYN, Feature Learning and more.

    Slides
  • Transfer Learning Overview & Finetuning Tutorial - Adrian Spataru

    Transfer learning is a machine learning method where a model trained for a domain is reused as the starting point for a model on a related domain. In this presentation, we explore different transfer learning techniques and explore the different type of pretrained models (Image, NLP) at our disposal. Furthermore, there will be a tutorial on finetuning for image recognition tasks.

    Slides Code
  • Hawkes Processes Tutorial - Tiago Santos

    Hawkes processes are a stochastic method for modeling discrete, inter-dependent events over continuous time. Problems like these can be found in a lot of domains. In finance, an event can represent a transaction on the stock market that influences future prices. In geophysics, an event can be an earthquake that is indicative of other earthquakes in the vicinity.

    Slides
  • Reinforcement Learning & Google DeepMind Talk - Anand Subramoney

    Reinforcement Learning Tutorial - Talk about the two main classes of learning algorithms (value-based learning and policy-based learning) with examples

    Google DeepMind Talk - Talk about following deep mind papers: • Mnih, V. et al. Human-level control through deep reinforcement learning. Nature 518, 529–533 (2015). • Mnih, V. et al. Asynchronous Methods for Deep Reinforcement Learning. arXiv:1602.01783 [cs] (2016).

    Slides Files
  • Dockerize Your Data Science Environment - Florian Geigl

    A Presentation on the open source Docker container used at Detego for all data science related tasks, followed by a live demo and some best practices.

    Slides Files
  • Predicting Instagram Likes - Adrian Spataru

    Can we predict likes in online social networks? What makes a post "attractive"? In this presentation, we will try to answer these questions.

    Slides
  • Machine Learning 101 with Scikit-Learn - Adrian Spataru

    Understand the basics of Machine Learning and build your first model.